Central Banks & Monetary Policy

Phase 7 · Week 58 · ~20 hours

Where it sits. Six people in Mumbai, or twelve in Washington, decide a single number. That number is the first number in the first line of every valuation, whether or not the person building the model realizes it. This is the second module of the macro phase. M7.01 built the machine from first principles: spending = money + credit; how commercial banks create money when they lend (the "loans create deposits" correction that endogenous money makes to the textbook multiplier); why savings equals investment; and Dalio's three forces, productivity and the short- and long-term debt cycles. This module puts the conductor on that machine. A central bank is the institution that sits at the apex of the money-and-credit system and tries to steer it: it is the banker to the commercial banks, the banker and debt-manager to the government, and the monetary authority that sets the price of the shortest, safest money in the economy. Credit expanding and contracting, the debt cycle turning: M7.01 described all of that as happening automatically, and the central bank is trying to modulate it, imperfectly, with a small set of tools and a long, uncertain delay before anything it does shows up in the real economy. This module teaches what those tools are and exactly how each one works. It also teaches the analyst's real skill, which is how to read a policy statement and know what the central bank is about to do to the discount rate that sits underneath every valuation you will ever build. The very next module (M7.03) takes the policy rate this module produces and follows it out the yield curve into inflation and asset prices; this one is where the rate is made.

Suggested hour budget. Core content: 8h · worked examples (the transmission trace, the QE balance-sheet mechanics, the statement read): 3h · practice set (trace-the-transmission, tool-matching, statement-reading): 4.5h · applied mini-project (read one live RBI and one live Fed statement side by side): 2.5h · flashcards + mastery check: 2h.

App link. This module feeds the Macro & Cycles branch and, specifically, its scenario-event engine (competency C8): the "rate hike / rate cut" event that fires mid-simulation and asks what do you do? is this module's transmission mechanism made playable. Four drills seed here: a tool-matching drill (a described action in, the correct tool + which central bank uses it out), a trace-the-transmission drill (a policy-rate move in, the ordered chain to inflation out, timed), a hawkish/dovish statement-reading drill (a statement excerpt in, the stance + likely next move out, wired to a bank of real RBI and Fed language), and a corridor/aggregates numeric drill (compute the corridor width, the money multiplier, an EMI change). It hands the policy rate forward to the valuation sandbox's discount-rate input, closing the loop the whole program is built on: macro sets the gravity, and gravity sets the price.

Learning objectives

By the end you can:

1. Explain what a central bank is and does from first principles: the three functions (banker to the banks; banker to and debt-manager for the government; monetary authority) and the lender-of-last-resort role. Say why a system of private money-creating banks needs an apex institution at all.
2. Compare the mandates of the US Federal Reserve (the dual mandate: maximum employment and price stability), the European Central Bank (price-stability primacy), and the Reserve Bank of India (flexible inflation targeting, a 4% CPI target with a ±2% tolerance band, dated as of mid-2026), and describe how the FOMC and the MPC are structured and actually vote.
3. Describe how a central bank sets "the" interest rate: the policy rate and the liquidity corridor (the repo rate and the SDF/MSF corridor in India; the fed funds target range and IORB/ON-RRP floor in the US). Explain what open-market operations do that a rate announcement alone does not.
4. State each tool's mechanism with an example: the policy rate, OMOs, QE/QT, the CRR and the (India-specific) SLR, forward guidance, and FX intervention. In particular, explain precisely what QE does and does not do, using the reserves-versus-broad-money distinction inherited from M7.01.
5. Define the money-supply aggregates M0/M1/M2/M3 (what each includes, India and US), and explain why base money ≠ broad money, tying the gap back to bank money-creation and forward to the money multiplier.
6. Trace the monetary-transmission mechanism from a policy-rate change through money-market rates → bank deposit and lending rates → the cost and availability of credit → borrowing, spending and investment → the output gap → inflation, across all the channels. Explain Friedman's "long and variable lags" and why they make policy genuinely hard.
7. Read a policy statement and the minutes like an analyst: decode hawkish versus dovish language, the stance, the vote split, and the Fed's Summary of Economic Projections and dot plot. Separate the market-moving signal from the boilerplate.
8. Argue the central-bank-independence debate evenhandedly, and explain why the policy rate is the "gravity" of every valuation: the bridge to M7.03 (rates and the yield curve) and back to M3.03 (the cost of capital).

Prerequisites & connections

Builds on. M7.01: this module is unintelligible without the money-creation mechanics. That commercial banks create broad money by lending (loans create deposits), that the central bank creates only base money (reserves + currency), and that the two are different things is the single most important idea carried forward, and it is the whole key to understanding QE. M7.01's short-term debt cycle is the thing monetary policy is trying to smooth. M3.01 gave you time value of money and discounting, and the reason a change in "the rate" changes every asset price is pure present-value arithmetic, and §4.10 cashes that in. M3.03 built the cost of capital: the risk-free rate that anchors CAPM and WACC is built on the policy rate and its expected path, so this module explains where the very first input of every valuation comes from. M3.05 covered reverse DCF and duration: "long-duration assets fall hardest when rates rise" is a sensitivity you already computed; here you learn what moves the rate. M5.01 covered banks: the CRR, SLR, CASA, NIM, and the CD ratio are the plumbing through which policy actually reaches the economy, and a bank's asset-versus-liability sensitivity (how fast loans reprice versus deposits) is the transmission mechanism seen from inside one firm. M0.03 gave you real versus nominal (Fisher): inflation is what monetary policy targets, and the real rate is what actually matters for behaviour.

Feeds forward. M7.03 (inflation, rates, and the yield curve) takes the policy rate produced here and follows it out the curve (the 10-year is roughly the average expected future policy rate plus a term premium), into the CPI/core/PCE/WPI measures the mandates target, and into the inversion signal. M7.04 (fiscal policy and sovereign debt): the fiscal–monetary interaction, deficit monetization, and fiscal dominance are the dark twin of central-bank independence (§4.9), and QE's blurring of the line between monetary and debt-management operations is where they meet. M7.05 (cycles, bubbles and crises): the lender-of-last-resort role (§4.1) is the central bank's crisis function, and every case study (2008, 2020) turns on what the central bank did and how fast. M7.06 (FX and the dollar system): FX intervention (§4.4) and the impossible trinity are previewed here and formalized there; the Fed's swap lines are lender-of-last-resort for the global dollar system. M7.07 (the dashboard capstone): the policy rate, its implied path, the balance sheet, and the money aggregates are core dashboard indicators, read as rate-of-change versus expectations. Competency C8, "understand the macro machine", is built substantially here: after this module you can read a central-bank statement and know what it means for the gravity under every price.

The one-sentence version of this module. A central bank sets the price and quantity of the safest, shortest money in the economy, and through a chain of bank rates, credit, and spending that works with a long and unpredictable lag, tries to keep inflation near a target (and, for some, employment near its maximum); learning to read what it is about to do, and why, is learning to read the single number that sits under every discount rate you will ever use.

Core teaching content

4.8 Reading a policy statement like an analyst

Decision day is the analyst's set-piece, and the skill is specific and learnable: the decision itself is usually the least informative part (it is often expected and pre-priced), and the signal is in the language, the projections, and the split. Here is what a trained reader actually extracts, in order of information content.

1. The decision vs. the expectation. A rate move only matters relative to what the market already priced. A "hike" that was fully expected can send yields down if the accompanying language is softer than feared; a "hold" can be violently hawkish if the guidance shifts. Always read the decision as a surprise (actual minus expected), never as a level — the market trades the second derivative. (This is M7's version of M2.05's "no number in isolation" and M6.02's Marks "what's priced in.")

2. The stance / guidance: the single most important line. Both central banks signal the direction of the next move in coded language. The RBI names an explicit stance, historically "accommodative," "neutral," or "withdrawal of accommodation". A change of stance (say, "withdrawal of accommodation" → "neutral") is a bigger signal than the rate decision, because it says the next moves have flipped direction. The Fed encodes guidance in sentences that are re-read word-for-word each meeting; the market watches for deletions and additions: dropping "the Committee anticipates further tightening," adding "data-dependent," changing "will be appropriate" to "may be appropriate." A one-word change is a headline.

3. Hawkish vs. dovish: the vocabulary you must own. These two words organize all central-bank reading:

• Hawkish = leaning toward tighter policy (more worried about inflation than growth): higher rates, faster QT, phrases like "inflation risks remain to the upside," "restrictive for longer," "not confident inflation is sustainably at target." A hawkish surprise → yields up, currency up, long-duration and rate-sensitive assets down.
• Dovish = leaning toward easier policy (more worried about growth/employment): lower rates, phrases like "risks are becoming balanced," "monitoring growth," "prepared to adjust." A dovish surprise → yields down, currency down, equities (especially long-duration growth) up.

A worked vocabulary drill lives in the practice set; the meta-skill is to read every sentence as tilting the scale toward hawk or dove relative to last meeting's sentence.

4. The dot plot and the projections (Fed) / the forecasts (RBI). The Fed's quarterly Summary of Economic Projections (SEP) includes the famous "dot plot": an anonymized scatter of each of the 19 FOMC participants' views of the appropriate fed funds rate at the end of each of the next few years and in the "longer run." Read it correctly, which most people don't: (a) the median dot is the headline, the committee's central view of the rate path; (b) the dispersion of the dots signals disagreement/uncertainty; (c) the "longer-run" dot is the committee's estimate of the neutral rate (r*), the rate that neither stimulates nor restrains, which tells you how restrictive current policy is (policy rate minus the longer-run dot = how much "restriction" is in the system); and (d), the professional's habit, compare the dots to what the market is pricing (from fed funds futures / overnight-index swaps). The gap between the dots and the market is the trade. When the Fed's dots show two cuts next year and the market prices four, someone is wrong, and the repricing when the gap closes is where money is made and lost. The RBI does not publish a dot plot, but it publishes explicit quarterly CPI and GDP growth projections, and the analyst does the same exercise: compare the RBI's inflation forecast path to the market's and to your own.

5. The vote split and the minutes. A divided vote (an FOMC dissent; a 4–2 MPC split) reveals the balance of the committee and foreshadows the next move: a lone dissenter for a hike is often the leading edge of a majority six months later. The minutes (Fed: three weeks later; RBI: ~14 days later) give the reasoning and the range of views, and are read for nuance the statement smoothed over, for which members are worried about what, and what would change their minds.

6. The press conference / Governor's statement. The Fed Chair's and RBI Governor's live remarks routinely move markets more than the statement, because the unscripted answers reveal the reaction function: what the central bank would do in scenarios not yet realized. The analyst listens for the conditionality: "what would make you cut faster?" and the answer is the map of the future.

The output of a good statement-read is not "they held" but a compact analyst's verdict: "Held at 5.5% as expected, but shifted the stance to neutral from withdrawal-of-accommodation and cut the FY inflation forecast by 30 bps — a dovish hold; the door is open to a cut at the next meeting, and the two-year yield should fall. Market was already ~half-pricing this, so the surprise is modest." That sentence is the skill.

4.10 Why it all matters: the rate is the gravity

Return, finally, to why a business analyst, not a bond trader, has just spent twenty hours on central banks. The answer is the most important single sentence connecting this phase to everything before it, and it is Warren Buffett's: "Interest rates are to asset prices what gravity is to the apple." Push rates down and the pull weakens, so prices drift up; raise them and everything gets dragged back toward earth. Here is the mechanism, in the language of the valuation phase you already own.

Every asset is worth the present value of its future cash flows (M3.01, M3.04). Present value means discounting, and the discount rate is built, layer by layer, on the risk-free rate (M3.03): the risk-free rate is the yield on government debt, which is anchored at the short end by the policy rate this module produces and set at the long end (the 10-year, which is what actually discounts a business's decade of cash flows) by the market's expectation of the average policy rate over that horizon plus a term premium (M7.03). So the chain is unbroken and direct: central bank sets the policy rate and its expected path → that sets the risk-free curve → that sets the discount rate in every CAPM and WACC → that sets the present value of every cash flow → that sets every asset price. The policy rate is quite literally the first number in the first line of every valuation, whether the person building the model realizes it or not.

Three consequences make this concrete and hand off to the next modules:

• Higher rates lower all asset prices, all else equal: the discount rate in the denominator rises, so present values fall. This is not a market opinion; it is arithmetic. When the Fed and RBI hiked sharply in 2022, the simultaneous fall in bonds, equities, real estate, and crypto was not five separate stories — it was one story: gravity increased.
• Long-duration assets fall hardest (the M3.05 duration point, now with its cause). A high-growth company whose value sits mostly in distant terminal cash flows is like a long-dated bond: its present value is acutely sensitive to the discount rate, so a rate rise hits it far harder than a stable, cash-now business (a utility, a consumer staple). This is the reason unprofitable-tech and richly-valued growth stocks were crushed in the 2022 hiking cycle while low-multiple value held up, a rotation you can now explain mechanically rather than narratively. The reverse is equally true: when the market anticipates cuts (dovish central bank), long-duration growth outperforms, because gravity is easing.
• The rate is the hurdle, so it governs capital allocation (M3.03, M3.07). When the risk-free rate is 1%, a project earning 6% clears the bar and capital floods into it; when the risk-free rate is 5%, the same project is value-destroying and capital retreats. The central bank, by setting the rate, is setting the hurdle rate for the entire economy's investment decisions. That is how monetary policy, through a thousand individual firms' capital-budgeting choices, becomes the investment cycle. A firm's WACC, the thing you spent M3.03 building, moves with the policy rate, so the same business is worth more or less depending on what six people in Mumbai or twelve in Washington decide.

That is why this module sits where it does. You cannot value a business without a discount rate; you cannot understand the discount rate without the risk-free rate; and you cannot understand the risk-free rate without the central bank. M7.03 now takes the policy rate out along the yield curve and into inflation and the recession signal; M3.03 is where the rate becomes a WACC; and every teardown you run for the rest of your life will, somewhere in it, contain an assumption that traces straight back to a central-bank statement you now know how to read.

Common mistakes & how experts think differently

1. "The central bank prints the money in my wallet / QE floods the economy with cash." The commonest and most consequential error. The central bank creates base money (reserves + currency); the money you spend is overwhelmingly broad money created by commercial-bank lending (M7.01, §4.6). QE creates reserves that are trapped in the banking system: it swaps bonds for reserves and (partly) deposits, changing the composition and price of assets, not primarily the quantity of spendable money. Experts ask "did this create new lending or new fiscal spending?", because that is what makes inflationary broad money, and are unsurprised that a decade of QE coincided with below-target inflation.

2. Confusing the policy rate with the rates that actually discount cash flows. The central bank sets one overnight rate. Your DCF uses the 10-year (M3.03/M7.03), which is set by the expected path of the policy rate plus a term premium. That is why US mortgage rates can fall while the Fed is still hiking. Beginners plug "the Fed's rate" into a long-horizon valuation; experts use the curve and know the policy rate is only its front anchor.

3. Reading the decision instead of the surprise. A hike can be bullish and a hold can be bearish. What moves markets is actual minus expected plus the change in guidance. Treating the headline number in isolation is the M2.05 "no number without a benchmark" error in macro clothing. The benchmark here is what was priced in (§4.8, and M6.02's second-level thinking).

4. Ignoring the lags: assuming policy "isn't working" or "already worked." Because the peak effect lands 12–24 months out and the lag is variable, the last three months of data tell you almost nothing about whether current policy is right. Pundits declare victory or failure on monthly prints; experts hold Friedman's lag in mind and read policy as steering-with-a-delay under uncertainty. The corollary: much of the effect of today's stance is still in the pipe, invisible.

5. Believing the money multiplier is a lever (the textbook-causation trap). The "central bank sets reserves, a fixed multiplier delivers deposits" story has the causation backwards and the stability wrong. Lending creates deposits and then reserves accommodate (endogenous money); the multiplier is an unstable ex-post ratio that collapsed after 2008. Watching M0 to predict M3 is a beginner's tell.

6. Treating a "dovish cut" and a "hawkish cut" as the same event. The action (a cut) is half the message; the guidance around it is the other half and often the bigger one. A cut paired with "this may be the last one for a while" is hawkish and can send yields up; a hold paired with a dovish stance-shift can send them down. Experts read the action and the words as one combined signal.

7. Forgetting that FX intervention is a domestic-liquidity operation. When the RBI sells dollars to support the rupee, it drains rupee liquidity and tightens domestic conditions unless it sterilizes. Analysts who track only the headline repo rate miss that heavy unsterilized FX intervention is silently moving the effective monetary stance — and the rate under every Indian valuation.

8. Taking central-bank independence for granted. Independence is operational, partial, and politically negotiated: the government sets the target and appoints members, and the norm can fray (India 2018). The market prices perceived independence: pressure to cut for political reasons shows up as a weaker currency and higher long yields, because credibility is the asset being spent. Ignoring this is how an EM analyst gets blindsided.

9. Symmetry blindness: assuming the problem is always inflation. Japan (1995–2024) fought deflation for a generation; the toolkit behaves completely differently at the zero bound ("pushing on a string"). An analyst who only knows the inflation-fighting playbook misreads a balance-sheet recession (M7.05) entirely.

How experts think differently, in one line: the amateur watches the rate decision; the professional watches the change in the expected path (dots vs market), the stance, the balance sheet, and broad-money/credit growth. Then, crucially, the professional remembers that whatever the central bank just did will not fully land for a year or two, so the job is to position for the pipe, not the print.

Worked examples

Worked Example 1: Trace the transmission: a 50-bp RBI repo cut reaches a home loan and a bank's margin (India)

The single most useful drill in the module: follow one rate decision from the MPC room to a household's bank account, showing every joint and its lag.

Setup. The MPC cuts the repo rate by 50 bps, from 5.50% to 5.00%. Trace the effect on (a) the corridor, (b) a household's EBLR-linked home loan, and (c) the lending bank's net interest margin in the following two quarters.

(a) The corridor moves immediately (days). The corridor is symmetric ±25 bps around the repo, so it slides down as a block: SDF 4.75% (floor), repo 5.00% (policy), MSF 5.25% (ceiling). The overnight WACR, arbitraged inside the corridor, falls to ~5.00% within a day or two; T-bill, CP, and CD rates follow within a week. Step 1 of transmission is essentially complete and reliable.

(b) The home loan reprices (within a quarter). Consider a ₹50 lakh, 20-year (240-month) floating-rate home loan whose rate is EBLR-linked at "repo + 350 bps spread," so it falls from 9.00% to 8.50% at the next reset (mandated within a quarter). Compute the EMI at each rate with the standard formula:

EMI = P × r × (1+r)^n / [(1+r)^n − 1], where P = principal, r = monthly rate (annual ÷ 12), n = number of months.

• At 9.00%: r = 0.09/12 = 0.0075; (1.0075)^240 ≈ 6.009. EMI = 5,000,000 × 0.0075 × 6.009 / (6.009 − 1) = 5,000,000 × 0.0450675 / 5.009 ≈ ₹44,986/month.
• At 8.50%: r = 0.085/12 = 0.0070833; (1.0070833)^240 ≈ 5.441. EMI = 5,000,000 × 0.0070833 × 5.441 / (5.441 − 1) = 5,000,000 × 0.038540 / 4.441 ≈ ₹43,391/month.
• The household's monthly saving: 44,986 − 43,391 = ~₹1,595/month, about ₹19,100 a year of freed cash flow (and, if the tenor is held constant, a large cut in total nominal interest over the life of the loan). That freed cash flow, multiplied across millions of floating-rate borrowers, is Step 4 (spending) beginning to work. Note the leak: a borrower on the old MCLR benchmark, or on a fixed-rate loan, sees the cut far more slowly or not at all. Transmission is not uniform.

(c) The bank's margin compresses first, then normalizes (the asset-sensitivity point, M5.01). The bank's EBLR-linked loan book reprices down almost immediately (within the quarter), but its term-deposit funding is locked at the old, higher rates until each deposit matures over the next 1–3 years. So in the two quarters after a cut, the bank's asset yields fall faster than its funding costs: NIM compresses (an asset-sensitive balance sheet is hurt by cuts and helped by hikes). Illustratively, if 60% of the loan book reprices down 50 bps this quarter but only 15% of deposits reprice, the near-term margin drag is real, which is why bank stocks often wobble on a rate cut even though a cut is "good for the economy." Over subsequent quarters, as deposits reprice down too, the margin recovers. The teaching point: the same event (a cut) is stimulative for the borrower and, transiently, a headwind for the lender's margin. An analyst must hold both, mapping the macro move onto the specific balance sheet (M5.01) it hits.

Timeline summary: corridor & money-market rates (days) → loan reprices (≤1 quarter) → bank NIM compresses then recovers (1–3 quarters) → household spending lifts (2–4 quarters) → output gap and inflation respond (12–24 months, variable). One decision; a two-year fuse.

Worked Example 2: The reserves-versus-money mechanics of QE, on balance sheets (US/global)

The example that inoculates you against the "money printing → hyperinflation" error for good, by doing the T-accounts.

Setup. The central bank conducts $10bn of QE. Trace the balance-sheet entries in two cases, buying from a bank and buying from a non-bank, and state the effect on base money and on broad money in each.

Case A: the central bank buys $10bn of bonds from a commercial bank.

• Base money (M0): +$10bn (reserves were created).
• Broad money (M1/M2/M3): unchanged. No household or firm has a single extra dollar of deposits; the bank merely swapped one asset (a bond) for another (reserves). This is the case that dominates QE, and it creates zero new spendable money. The reserves sit at the central bank.

Case B: the central bank buys $10bn of bonds from a non-bank (a pension fund). The fund has no reserve account, so payment routes through the fund's bank:

• Base money (M0): +$10bn (reserves again).
• Broad money: +$10bn, since the pension fund now holds a deposit where it held a bond. So QE bought from non-banks does raise broad money.

But read Case B correctly. This is the whole point. The pension fund is no richer: it swapped a $10bn bond for a $10bn deposit of identical value. Nothing forces it to spend the deposit; most likely it uses the cash to buy another asset (a corporate bond, equities: the portfolio-rebalancing channel), pushing those prices up and yields down, which is precisely the financial-conditions effect QE is designed to produce. Inflation only follows if those lower yields and higher asset prices induce new borrowing, investment, and spending in the real economy, a behavioural "if," not a mechanical certainty. Conclusion: QE reliably inflates asset prices (its job) and only contingently inflates consumer prices (via induced spending). The 2010s (QE, low CPI) versus 2021 (QE plus trillions of fiscal transfers deposited straight into spending accounts plus supply shocks, high CPI) is the natural experiment that proves reserves ≠ spending, and that the inflationary fuel was the fiscal broad money and the supply constraints, not the reserves per se.

The multiplier footnote. Across the whole 2008–2021 US QE program, the monetary base (M0) rose roughly seven-fold while M2 (broad money) rose far less than proportionally, so the money multiplier collapsed (from ~9 pre-crisis toward ~3–4). If the textbook "fixed multiplier" were real, that base expansion would have produced hyperinflation; it did not, because reserves are not broad money and banks do not "lend out reserves." The collapsed multiplier is the endogenous-money view winning the argument in the data.

Worked Example 3: Read the statement: decode a synthetic FOMC decision and dot plot

Operationalizes §4.8: turn a wall of central-bank prose and a scatter of dots into a one-sentence analyst verdict and a trade.

Setup (synthetic, illustrative). The FOMC holds the target range at 3.75–4.00% (as widely expected). The statement changes two sentences versus the prior meeting: it deletes "the Committee anticipates that additional policy firming may be appropriate" and adds "the Committee will assess incoming data in determining the extent to which further adjustments may be appropriate." The new SEP dot plot shows a median dot of 3.4% for end-next-year (implying ~two 25-bp cuts from here) and a longer-run dot of 3.0%. Fed funds futures, meanwhile, price the end-next-year rate at 2.9% (implying ~four cuts). Produce the analyst read.

Step 1: decision vs. expectation. A hold, fully expected → the decision itself carries almost no information. Move to the language.

Step 2: the guidance change. Deleting "additional firming may be appropriate" removes the hiking bias; adding "assess incoming data… extent of further adjustments" is classic data-dependent, we're-done-hiking language. This is a dovish shift in guidance: the bigger signal than the hold.

Step 3: the dots. Median end-year dot 3.4% vs. the current 3.875% midpoint → the committee's own central view is ~two cuts next year: consistent with the dovish guidance. The longer-run dot of 3.0% is the committee's neutral-rate (r*) estimate, so current policy (3.875%) is ~0.9% restrictive, still leaning against the economy, with room to ease toward neutral.

Step 4: dots vs. market (the trade). Here is the professional's move: the market (2.9%) is pricing two more cuts than the median dot (3.4%). The market is more dovish than the Fed. That gap is the trade: either incoming data soften and the Fed migrates down toward the market (dots fall, bullish for duration and growth equities), or the economy stays firm and the market capitulates up toward the dots (front-end yields rise, painful for rate-sensitive positions). An analyst does not assume the Fed is right and the market wrong (or vice versa); the analyst identifies the gap as the key uncertainty and watches the data (jobs, core PCE) that will resolve it.

Step 5: the one-sentence verdict. "A widely-expected hold, but a clearly dovish shift: the hiking bias was dropped and the language went data-dependent, and the dots imply the Fed thinks it's roughly done and ~two cuts from neutral next year. The market is pricing more easing than the dots, so the live question is who converges to whom — watch core PCE and payrolls; a soft print pulls the dots to the market (risk-on, duration rallies), a hot one pulls the market to the dots (front-end sells off)." That paragraph, not "the Fed held", is the deliverable, and it is exactly what the applied mini-project asks you to produce on a live statement.

Worked Example 4: Is policy tight or loose? The Taylor rule and the real repo rate (India)

Two back-of-envelope gauges that turn "the repo rate is 5.25%" into "policy is on the easy side of neutral", the judgment §4.3 and §4.10 need but never quantify.

Setup. As of mid-2026 the RBI's repo rate is 5.25% and the inflation target is 4% headline CPI (both public; check at rbi.org.in). Take three illustrative estimates an analyst must supply and mark synthetic: a neutral real rate r* = 1.5%, current CPI inflation π = 4.5%, and an output gap of +0.5% (output just above potential). Is the 5.25% repo tight or loose?

Step 1: the real policy rate (the Fisher lens). Behaviour responds to the real rate, not the nominal headline (M0.03). Real repo ≈ nominal − inflation = 5.25 − 4.5 = +0.75% (approximation); exact Fisher (1.0525 / 1.045) − 1 = +0.72%. After inflation, money costs only about three-quarters of a percent, barely positive in real terms, though the nominal 5.25% "sounds" high.

Step 2: compare the real rate to neutral. The neutral (real) rate r* is the real rate that neither stimulates nor restrains. With r* ≈ 1.5% and the real repo at ≈ 0.72%, the real rate sits ~0.8 pp below neutral → policy is accommodative (pushing on the economy, not leaning against it). Note the trap this kills: a positive real rate is not automatically "tight"; tightness is measured against neutral, not against zero.

Step 3: the Taylor-rule cross-check. John Taylor's rule prescribes a nominal rate from the same ingredients: i = r\* + π + 0.5(π − π\*) + 0.5(output gap). Plugging in: 1.5 + 4.5 + 0.5(4.5 − 4.0) + 0.5(0.5) = 1.5 + 4.5 + 0.25 + 0.25 = 6.5%. The rule "wants" 6.5%; the actual repo is 5.25%. The rule sits 125 bps above actual, again saying policy is easier than a mechanical benchmark would set.

Step 4: read it, don't obey it. Both gauges agree: as of mid-2026 the RBI is running policy on the accommodative side of neutral, consistent with having cut from 6.5% through 2025 to support growth while inflation sits near target, which is exactly the flexible in flexible inflation targeting (§4.2). But the Taylor rule is a benchmark, not a mandate: no central bank obeys it mechanically; r* and the output gap are unobservable estimates (nudge them and the "prescription" moves), and the RBI weighs growth, the rupee, and financial stability besides. The teaching point: never say "the repo is 5.25%" and stop. Convert it to a real rate, compare it to neutral, and sanity-check with a Taylor rule, and you can state whether the single most important number under every Indian valuation is helping or fighting the economy right now.

Worked Example 5: One rate move, repriced through a 10-year bond (US $)

Puts a number on "the rate is gravity" (§4.10) using the instrument that actually anchors the discount rate, and shows why analysts quote duration, not just the yield change.

Setup. A 10-year US Treasury note (synthetic), $1,000 face, 4% annual coupon, initially priced at par because it yields 4%. The market's expectation of the Fed's rate path shifts up and the yield rises to 5% (a 100-bp move). What happens to the price, and to anything discounted off it?

Step 1: price it at each yield. Discount every cash flow (ten $40 coupons + the $1,000 face) at the yield:

• At 4%: price = $1,000.00 (coupon = yield → par).
• At 5%: Σ 40 / 1.05^t (t = 1..10) + 1000 / 1.05^10 = 308.87 + 613.91 = $922.78.

Step 2: the percentage hit. (922.78 − 1000) / 1000 = −7.72%. A one-percentage-point rise in the yield knocked nearly 8% off the price, with the coupon and maturity completely unchanged. Nothing about the bond changed; only the discount rate did.

Step 3: the duration shortcut (and its error). Analysts approximate this with modified duration: %ΔP ≈ −ModDur × Δy. This bond's modified duration at 4% is ≈ 8.11, so the linear estimate is −8.11 × 0.01 = −8.11%. The actual fall was −7.72%, so duration overstated the loss by ~0.4 pp. That gap is convexity: the price–yield curve bends, so duration (a straight line) overpredicts the loss on a rate rise and underpredicts the gain on a rate fall. For small moves duration is close; for big moves you add the convexity correction.

Step 4: the bridge to every valuation. This 10-year yield is not a bond-trader's toy. It is the risk-free rate that anchors the discount rate in every CAPM and WACC (M3.03), and its level is the expected average policy rate plus a term premium (M7.03). So when a hawkish shift lifts the 10-year 100 bps, it mechanically re-rates a par bond down ~7.7%, and a long-duration equity, whose cash flows sit even further out, falls more (P12, P17). This is Buffett's gravity (§4.10) with the arithmetic filled in: the central bank moves the front of the curve, expectations move the 10-year, and the 10-year moves the present value of everything. The teaching point: the same 100-bp move is trivial for a 1-year bill and brutal for a 10-year bond or a growth stock: duration is the multiplier that turns a rate change into a price change, which is why "rates up" is never one story across assets.

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. Compute money to the nearest sensible unit.

P1 (guided — tool-matching). For each described action, name the tool and say which central bank uses it as a live, routine tool (RBI, Fed, both): (a) obliging banks to hold 4% of deposits, non-interest-bearing, at the central bank; (b) buying $60bn of long-dated government bonds per month to lower long yields at the zero bound; (c) selling US dollars from reserves to slow a currency's fall; (d) requiring banks to hold 18% of deposits in government securities; (e) announcing "rates will stay at present levels at least through end-2025."

Solution. (a) Reserve requirement: the CRR; a live tool for the RBI, effectively dead for the Fed (which set reserve requirements to zero in March 2020). (b) Quantitative easing (QE); used by both in crises (the Fed massively in 2008/2020), though not India's day-to-day tool. (c) FX intervention; routine for the RBI (managed float), essentially never for the Fed (free float). (d) The SLR (Statutory Liquidity Ratio), RBI-specific, no Fed equivalent. (e) Forward guidance (calendar-based); both use it, the Fed most explicitly.

P2 (guided — the corridor). The RBI's repo rate is 6.00%, with a symmetric ±25 bp corridor. (a) State the SDF and MSF rates and name which is the floor and which the ceiling. (b) What is the corridor's width in bps? (c) If the RBI wants to nudge the overnight rate up toward the repo rate without changing any of these three rates, what does it do?

Solution. (a) SDF = 5.75% (the floor: the deposit facility banks earn on spare reserves) and MSF = 6.25% (the ceiling: the lending facility banks pay when short). The repo (6.00%) sits in the middle. (b) Width = MSF − SDF = 6.25 − 5.75 = 50 bps. (c) It drains liquidity via a variable-rate reverse-repo / OMO sale, reducing the quantity of surplus reserves so the overnight rate drifts up off the floor toward the repo rate. This is a real (marginal) tightening without any change in the headline rate — the move the casual reader misses.

P3 (guided — hawkish or dovish?). Classify each statement change as hawkish, dovish, or neutral, and give a one-line reason: (a) deleting "further tightening may be appropriate"; (b) raising the projected inflation forecast and adding "risks are to the upside"; (c) shifting the stance from "withdrawal of accommodation" to "neutral"; (d) a 4–2 vote to hold, with the two dissenters wanting a hike.

Solution. (a) Dovish: removing the hiking bias signals the tightening cycle is ending. (b) Hawkish: a higher inflation forecast with upside-risk language leans toward tighter-for-longer. (c) Dovish (a dovish shift): moving from "withdrawal of accommodation" to "neutral" says the next moves are no longer biased toward tightening; a stance change often out-signals the rate decision. (d) Hawkish at the margin: two members voting for a hike reveals a hawkish minority that could become a majority; the split foreshadows the next move even though the decision was to hold.

P4 (guided — the aggregates and the multiplier). (a) Classify each into the narrowest aggregate it belongs to (M0, M1, or M3, India definitions): (i) ₹2,000 notes with the public; (ii) a bank's reserve balance at the RBI; (iii) money in your current (checking) account; (iv) a 3-year fixed deposit; (v) CRR balances at the RBI. (b) If reserve money (M0) is ₹48 lakh crore and broad money (M3) is ₹264 lakh crore, compute the money multiplier. (c) What does a multiplier of that size tell you about who creates most of the money?

Solution. (a) (i) Currency with the public → M1 (and hence M3). (ii) Bankers' deposits with the RBI → M0 (base money, not M1/M3; reserves are not the public's money). (iii) Demand deposit → M1. (iv) Time deposit → M3 (it is not in M1). (v) CRR balances are bankers' deposits at the RBI → M0. (b) Multiplier = M3 ÷ M0 = 264 ÷ 48 = 5.5×. (c) It tells you that for every ₹1 of base money the central bank created, the banking system supports ₹5.5 of broad money, i.e., the great majority of money in the economy was created by commercial banks lending (loans create deposits), not by the central bank "printing." The multiplier is the result of that credit creation, not a lever the central bank pulls.

P5 (independent — order the transmission chain). Put these links of the transmission mechanism in the correct order, and label the approximate lag of each: (i) household spending on cars and homes rises; (ii) the overnight money-market rate falls to the new repo rate; (iii) inflation edges up; (iv) EBLR-linked loan rates reprice lower; (v) banks' credit growth accelerates.

Solution. Order: (ii) → (iv) → (v) → (i) → (iii). (ii) Overnight rate falls to the new policy rate: days (fast, reliable). (iv) EBLR-linked loans reprice down: within a quarter (mandated reset). (v) Cheaper, more-available credit → credit growth accelerates: one to two quarters (price and the bank-lending/availability channel). (i) Households borrow and spend more on interest-sensitive durables: two to four quarters. (iii) Higher demand closes the output gap and, if demand outruns supply, inflation rises: 12–24 months, variable (Friedman's long-and-variable lag). The chain fires like a slow fuse: the front is quick and controllable, the back is slow, leaky, and uncertain.

P6 (independent — QE reserves vs money). The Fed buys $20bn of Treasuries via QE: $15bn from commercial banks and $5bn from pension funds. (a) By how much does base money (M0) rise? (b) By how much does broad money (deposits) rise, and from which portion? (c) Explain why this operation need not be inflationary, in one or two sentences.

Solution. (a) Base money rises by the full $20bn: the Fed credited $20bn of new reserves regardless of who sold. (b) Broad money rises by $5bn: only the portion bought from non-banks (pension funds), who now hold a deposit where they held a bond; the $15bn bought from banks created reserves but no new deposit (the banks just swapped a bond for reserves). (c) QE swaps assets in private hands (bond → deposit) without making anyone richer or forcing anyone to spend; the reserves are trapped in the banking system, and the $5bn of new deposits will most likely be reinvested in other assets (lifting asset prices, QE's real job) rather than spent on goods. Inflation follows only if the lower yields QE produces induce new borrowing and spending, a behavioural "if," which is why a decade of QE coincided with below-target inflation.

P7 (independent — the transmission leak). After a 50-bp repo cut, three borrowers at the same bank see very different outcomes: Borrower A (new EBLR-linked home loan) gets the full 50 bp cut within a quarter; Borrower B (older MCLR-linked loan) sees ~15 bps after two quarters; Borrower C (fixed-rate loan) sees nothing. (a) Explain the mechanism behind each. (b) What does this reveal about "transmission," and why did the RBI mandate EBLR in 2019? (c) Name one reason the deposit side transmits even more slowly.

Solution. (a) A is contractually linked to the external benchmark (the repo rate itself), so it reprices automatically and fully at the mandated reset. B is on the bank's internal MCLR (marginal cost of funds-based lending rate), which the bank moves slowly and only partially, at its discretion, as its own funding costs drift, the sticky, opaque legacy mechanism. C is fixed: its rate is contractually unchanged for the term, so policy does not reach it at all. (b) It reveals that transmission is leaky and non-uniform: the same policy move reaches different borrowers at very different speeds and magnitudes, so "the RBI cut 50 bps" does not mean the economy's borrowing cost fell 50 bps. The RBI mandated EBLR in October 2019 precisely to plug the leak: banks had been quick to pass on hikes but slow to pass on cuts via MCLR, so linking retail/MSME floating loans to an external benchmark forced faster, fuller pass-through. (c) Term deposits are locked at their contracted rate until maturity, so a bank's funding cost only falls as old deposits roll off over 1–3 years, which is why a cut compresses bank margins in the short run (assets reprice down before liabilities do).

P8 (independent — dots vs market). The Fed's median dot for end-next-year is 3.6%; fed funds futures price 3.1%; the current target midpoint is 4.125%; the longer-run dot is 3.0%. (a) How many 25-bp cuts does each of the Fed and the market imply by end-next-year? (b) Who is more dovish? (c) How restrictive is current policy, and where is "the trade"?

Solution. (a) Fed (dots): 4.125% → 3.6% = 0.525% ≈ ~2 cuts. Market: 4.125% → 3.1% = 1.025% ≈ ~4 cuts. (b) The market is more dovish: it prices twice the easing the Fed's own median projects. (c) Current policy is restrictive by ~1.125% (4.125% midpoint minus the 3.0% neutral/longer-run dot), so there is room to ease toward neutral. The trade is the ~2-cut gap between the dots and the market: it resolves either by the Fed migrating down to the market (if data soften, bullish duration/growth) or by the market capitulating up to the dots (if data stay firm, front-end yields rise). The analyst flags the gap as the key uncertainty and watches core PCE and payrolls to see who converges to whom.

P9 (independent — FX intervention and sterilization). The rupee is falling sharply and the RBI sells $5bn from reserves to slow it. (a) What happens to domestic rupee liquidity, and is this an easing or a tightening of domestic conditions? (b) If the RBI does not want that side effect, what offsetting operation does it run, and what is this called? (c) Connect the episode to the "impossible trinity" in one line.

Solution. (a) Selling dollars means the RBI receives rupees from banks and extinguishes them, so domestic rupee liquidity is drained, an inadvertent tightening of domestic monetary conditions (fewer reserves in the system, upward pressure on the overnight rate). (b) To neutralize the liquidity effect, the RBI simultaneously injects rupees back via an OMO purchase (or repo) of government securities, buying the drained rupees back into the system. This is sterilized intervention (affect the exchange rate without moving domestic liquidity). (c) The episode is the impossible trinity in action: with open-ish capital flows, defending the rupee's level would force the RBI to surrender control of domestic monetary policy, so India instead runs a managed float, smoothing volatility with reserves while letting the rupee move and keeping monetary independence.

P10 (exam — 12 min — the long and variable lags). A commentator says: "The central bank hiked 18 months ago and inflation is only now falling, while unemployment has just started to rise — clearly they hiked too much and are causing a recession." (a) What single concept must you invoke to assess this claim? (b) Explain why "wait until you see inflation fall, then stop hiking" is itself a way to over-tighten. (c) What does this imply about how the central bank must set policy, and what should the analyst not conclude from the last three months of data?

Solution. (a) Friedman's long and variable lags: the peak effect of a rate change on the real economy and inflation lands roughly 12–24 months later, and the exact lag is unpredictable. (b) Because the medicine already administered keeps working after you stop, a central bank that keeps hiking until it sees inflation fall has almost certainly put too much tightening in the pipe. The hikes of the last year are still working their way through when inflation finally turns, so "stop when you see it" guarantees over-shooting into a downturn. This is the standard mechanism by which central banks cause the recessions that end inflations, and it is a genuine (not obviously avoidable) dilemma, not proof of incompetence. (c) Policy must be forward-looking: set on forecasts of inflation and growth 1–2 years out, not on today's print (which reflects decisions made 1–2 years ago). The analyst should not conclude from three months of data that policy "already worked" or "did nothing". Much of the current stance's effect is still in the pipe and invisible, so recent prints are near-useless for judging whether policy is correctly calibrated.

P11 (exam — 12 min — mandates compared). A growth scare hits: unemployment is rising and inflation is at target (say 2%). (a) How does the dual-mandate Fed's likely reaction differ from the price-stability-primacy ECB's? (b) How does the RBI's flexible inflation-targeting framework sit between them? (c) Why does the choice of inflation measure and target (PCE vs headline CPI; who sets it) matter for how each behaves?

Solution. (a) The Fed, holding employment as a co-equal objective, can ease on the rising-unemployment leg even with inflation at target: a growth scare is a live reason to cut. The ECB, whose price-stability objective is primary and employment only "without prejudice" to it, is slower to ease on growth grounds alone with inflation at target. Its constitution says price stability comes first, so it needs inflation itself to be falling to justify easing. (b) The RBI is in between but closer to the ECB: inflation (4% CPI) is the primary target, growth a secondary, flexible consideration, so it can weigh the growth scare and "look through" temporary shocks, but growth is not a co-equal mandate as at the Fed; it would ease more readily than a pure inflation-targeter but less automatically than the Fed. (c) The measure and its setter shape behaviour: the Fed targets PCE (which runs a few tenths below CPI, so the same "2%" is a slightly easier bar than a CPI 2%); the RBI targets headline CPI including food and fuel, so it reacts to volatile food/energy shocks that a core-targeter would look through. And because the government sets the RBI's target (every five years) and appoints half the MPC, the framework is more politically embedded, which matters for the independence read (§4.9).

P12 (exam — 12 min — the rate is gravity). Two assets each promise a single ₹100 cash flow: Asset S in 1 year, Asset L in 30 years. The discount rate rises from 5% to 8% (a hiking cycle). (a) Compute each asset's present value at both rates. (b) Compute each asset's percentage change in value. (c) State the general principle and connect it to why unprofitable-growth stocks were crushed in 2022 while value held up.

Solution. (a) Asset S (1-yr): PV at 5% = 100/1.05 = ₹95.24; at 8% = 100/1.08 = ₹92.59. Asset L (30-yr): PV at 5% = 100/1.05³⁰ = 100/4.3219 = ₹23.14; at 8% = 100/1.08³⁰ = 100/10.0627 = ₹9.94. (b) Asset S: (92.59 − 95.24)/95.24 = −2.8%. Asset L: (9.94 − 23.14)/23.14 = −57.1%. (c) The general principle: the more distant (longer-duration) an asset's cash flows, the more its present value falls when the discount rate rises: the same rate move barely dents the near cash flow (−2.8%) but more than halves the distant one (−57%). This is the mechanism behind 2022: unprofitable-growth stocks are like Asset L, since nearly all their value sits in far-future ("terminal") cash flows, so when the Fed and RBI hiked, the rising discount rate crushed them, while cash-generative "value" businesses (short-duration, cash now, like Asset S) held up. Same gravity, very different sensitivity — exactly the duration point from M3.05, now with its macro cause.

P13 (guided — the real policy rate). The repo rate is 5.25% and expected CPI inflation is 4.5% (as of mid-2026). (a) Compute the real policy rate by the quick approximation. (b) Compute it exactly (Fisher). (c) If the neutral real rate is estimated at ~1.5%, is policy accommodative or restrictive, and why?

Solution. (a) Approximation: real ≈ nominal − inflation = 5.25 − 4.5 = +0.75%. (b) Exact Fisher: (1 + 0.0525)/(1 + 0.045) − 1 = 1.0525/1.045 − 1 = +0.72% (the approximation slightly overstates it; the two diverge as rates rise). (c) The real repo (~0.72%) is below the ~1.5% neutral real rate, so policy is accommodative: even though the real rate is positive, tightness is measured relative to neutral, not against zero, and 0.72% < 1.5% means the setting is still pushing on the economy rather than restraining it. (The trap: calling any positive real rate "tight.")

P14 (guided — the Taylor rule). Use i = r* + π + 0.5(π − π*) + 0.5(output gap) with a neutral real rate r* = 2%, an inflation target π* = 4%, current inflation π = 6%, and an output gap of 0 (economy at potential). (a) Compute the prescribed nominal policy rate. (b) If the actual repo rate is 6.5%, what does the rule imply about the stance? (c) Give one reason a flexible-inflation-targeting central bank might not follow the rule here.

Solution. (a) i = 2 + 6 + 0.5(6 − 4) + 0.5(0) = 2 + 6 + 1 + 0 = 9.0%. (b) The rule prescribes 9.0% versus an actual 6.5%, so it says the actual setting is 250 bps looser than the benchmark, because inflation (6%) is a full 2 pp above target. (c) A flexible inflation-targeter (the RBI, §4.2) might judge the 6% inflation a transient food/supply shock to look through rather than demand to crush; it also weighs the growth outlook and financial stability, and it knows r* is only an estimate. The Taylor rule is a disciplining benchmark, not an autopilot. (The trap: forgetting to add the level of current inflation π, a very common slip that wrongly yields ~3% instead of 9%.)

P15 (independent — a rate move through a bond). A 5-year US Treasury note (synthetic), $1,000 face, 3% annual coupon, is priced at par at a 3% yield. The yield rises to 4%. (a) Its price at 3%? (b) Its price at 4%? (c) The percentage change? (d) Would a 10-year note with the same coupon fall more or less for the same 1-pp yield rise, and why?

Solution. (a) At a 3% yield with a 3% coupon the bond is at par: $1,000.00. (b) At 4%: Σ 30/1.04^t (t = 1..5) + 1000/1.04^5 = 133.55 + 821.93 = $955.48. (c) (955.48 − 1000)/1000 = −4.45%. (d) More: a 10-year note has higher duration (its cash flows are more distant, so more of its value is exposed to discounting), so the same 1-pp yield rise cuts its price roughly twice as much (~−7.7%, Worked Example 5). Longer maturity → longer duration → bigger price move for a given rate change. (The trap: assuming the same yield change produces the same price change regardless of maturity; duration, not the yield move alone, sets the damage.)

P16 (independent — rate pass-through to a bank's NIM). A bank (synthetic) has ₹1,000 crore of interest-earning assets yielding 9%, funded by ₹1,000 crore of deposits costing 5% (ignore equity funding for simplicity). The RBI hikes 100 bps. In the first year, 70% of the assets reprice up by the full 100 bps but only 30% of the deposits do (the bank is asset-sensitive). (a) Net interest income and NIM before the hike? (b) After? (c) What happens to NIM, and what would a cut do?

Solution. (a) Before: NII = 1,000 × (9% − 5%) = ₹40 crore; NIM = 40 ÷ 1,000 = 4.0%. (b) New asset yield = 9% + 0.70 × 1% = 9.70%; new funding cost = 5% + 0.30 × 1% = 5.30%; NII = 1,000 × (9.70% − 5.30%) = ₹44 crore; NIM = 4.40%. (c) NIM expands by 40 bps: an asset-sensitive bank's assets reprice up faster than its funding, so a hike helps its margin in the near term. A cut does the reverse: assets reprice down faster than deposits, compressing NIM, exactly the drag on the lending bank in Worked Example 1, seen from the other direction. (The trap: "higher rates always hurt banks"; the sign depends on asset- versus liability-sensitivity, M5.01.)

P17 (independent — gravity: growth versus value). Two stocks (synthetic) are valued as growing perpetuities with next-year cash flow $2. Stock G grows at g = 4%; Stock V does not grow (g = 0%). The policy-driven discount rate rises from 8% to 9%. (a) Value each at 8% and 9% (V = CF₁ ÷ (r − g)). (b) The percentage change for each. (c) Which is the "longer-duration" asset, and how does this explain the 2022 growth-stock rout (§4.10)?

Solution. (a) Stock G: at 8%, 2 ÷ (0.08 − 0.04) = 2 ÷ 0.04 = $50.00; at 9%, 2 ÷ (0.09 − 0.04) = 2 ÷ 0.05 = $40.00. Stock V: at 8%, 2 ÷ 0.08 = $25.00; at 9%, 2 ÷ 0.09 = $22.22. (b) Stock G: −20.0% [(40 − 50) ÷ 50]; Stock V: −11.1% [(22.22 − 25) ÷ 25]. (c) Stock G is the longer-duration asset: more of its value sits in far-future cash flows (the small r − g denominator makes it acutely rate-sensitive), so the same 1-pp rise in the discount rate cuts it nearly twice as hard. That is the mechanism behind 2022: as the Fed and RBI hiked, the rising discount rate crushed long-duration growth stocks while shorter-duration "value" held up (P12, Worked Example 5). (The trap: dividing the change by the new value, or forgetting that the same rate move hits the two denominators very differently.)

P18 (exam — 12 min — is policy tight or loose, and where next?). On a synthetic decision day: repo 6.00%, expected CPI 5.0%, neutral real rate r* = 1.5%, target π* = 4%, output gap +1.0%. The MPC also shifts its stance to "neutral" from "withdrawal of accommodation" and cuts its inflation forecast. (a) Compute the real policy rate (exact Fisher). (b) Compute the Taylor-rule prescribed rate. (c) Relative to each benchmark, is the 6.00% repo easy or tight? (d) Classify the stance shift as hawkish or dovish and state its likely effect on the 2-year G-sec yield.

Solution. (a) Real repo = (1.06 ÷ 1.05) − 1 = +0.95%. (b) Taylor = 1.5 + 5.0 + 0.5(5.0 − 4.0) + 0.5(1.0) = 1.5 + 5.0 + 0.5 + 0.5 = 7.5%. (c) By both gauges the 6.00% repo is on the easy side: the real repo (0.95%) is below the 1.5% neutral real rate (accommodative), and the Taylor rule (7.5%) sits 150 bps above the actual repo (the mechanical benchmark "wants" tighter). (d) The stance shift from "withdrawal of accommodation" to "neutral" is a dovish shift: it says the next moves are no longer biased toward tightening, and pairing it with a lower inflation forecast reinforces the dovish read; the 2-year yield should fall (it prices the expected policy path, which just tilted toward cuts). Bottom line: a dovish hold that leaves policy on the accommodative side of neutral, with the door open to a cut. (The trap: reading the unchanged 6.00% as "no news"; the stance change and forecast cut are the signal, §4.8.)

Applied mini-project: read a live RBI and a live Fed decision, side by side

Task. Pull the most recent RBI Monetary Policy Statement and the most recent FOMC statement, and write a two-page comparative "decision read" that turns each wall of prose into an analyst's verdict, and then contrasts the two central banks' current postures. This is the exact skill §4.8 teaches and the certification-phase teardowns will assume you own; do it on live documents so you are reading today's actual language, not a frozen example.

Data to pull (all free, all primary):

• RBI: the latest Monetary Policy Statement and the MPC resolution (rbi.org.in → Press Releases / Monetary Policy), the Governor's statement, and, if released, the latest MPC minutes (~14 days after) and the Monetary Policy Report. Note the repo rate, the SDF/MSF corridor, the CRR/SLR, the stance, the RBI's CPI and GDP projections, and the vote split.
• Fed: the latest FOMC statement, the implementation note, the SEP + dot plot if it was a projection meeting (federalreserve.gov → Monetary Policy → FOMC), and the minutes (~3 weeks after). Note the target range, the guidance language, the median and longer-run dots, and any dissents.
• Market pricing: for the Fed, eyeball the market-implied path (the CME FedWatch tool, free, or overnight-index-swap commentary) to compare against the dots. For India, note the 10-year G-sec yield and any commentary on what the market expects next.
• Data context: the latest US core PCE and India headline CPI prints (FRED; MOSPI) so you can judge each decision against its own target.

Required contents.

1. Two one-paragraph "decision reads" (one RBI, one Fed), each following the §4.8 output format: decision vs. expectation → the stance/guidance change → hawkish/dovish verdict → the projections/dots → the vote split → your one-sentence bottom line and its implication for yields.
2. The dots-vs-market gap (Fed) and the RBI-forecast-vs-market gap: state where the central bank's implied path differs from the market's, and name the data that would resolve it.
3. The side-by-side contrast: are the two central banks in the same phase of their cycle (both easing? both on hold? diverging)? Explain why their postures differ, using the mandate differences (§4.2) and their different inflation/growth positions.
4. The valuation bridge (one paragraph): state, concretely, what each posture implies for the discount rate an analyst should use for a company in that economy right now, and therefore for long-duration vs short-duration equities there (§4.10).

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

Case Lab: run case CL-7.2 (Decision Day: One Repo Cut, Traced to a Valuation) in the app's Cases tab for this module, a synthetic MPC decision walked end to end: read the stance shift, slide the corridor, reprice a home loan and a bank's NIM, then watch the discount rate move under a sample company's valuation — the transmission chain of §4.7 and the gravity of §4.10 made playable.

Reading & resources

• Ray Dalio, How the Economic Machine Works: the free 30-minute video (YouTube) and the essay. The best possible intuition for money vs credit and the debt cycle within which central banks operate; watch it before this module if you haven't. [Free] [Beginner]
• Ben Bernanke, 21st Century Monetary Policy (2022): the definitive modern account of how the Fed's toolkit evolved (rates → QE → guidance → the 2020 framework), by the chair who built much of it. The single best book for this module. [Paid] [Intermediate]
• RBI: the Monetary Policy Framework page, MPC statements, and the Monetary Policy Report (rbi.org.in): your permanent primary source for the Indian target, the repo/SDF/MSF corridor, CRR/SLR, and the MPC's reasoning. Read two consecutive statements to see what language changed. [Free] [Beginner→Intermediate]
• Federal Reserve: FOMC statements, minutes, and the SEP/dot plot (federalreserve.gov): the primary source for US policy; pair with the CME FedWatch tool (free) to see the market-implied path against the dots. Read one statement with the prior meeting's open beside it and diff the sentences. [Free] [Beginner→Intermediate]
• FRED (fred.stlouisfed.org): pull and plot the series yourself: FEDFUNDS/DFF (fed funds), WALCL (Fed balance sheet; see QE and QT), M1SL/M2SL (money stock), BOGMBASE (monetary base; watch the multiplier collapse after 2008), WM2NS. The habit of plotting the series is worth more than any summary. [Free] [Beginner]
• RBI DBIE (Database on Indian Economy), the Indian analogue: reserve money (M0), M3, the corridor rates, and bank-credit growth. For the dashboard you build in M7.07. [Free] [Intermediate]
• Walter Bagehot, Lombard Street (1873): chs. on the lender of last resort; astonishingly, still the clearest statement of "lend freely, against good collateral, at a penalty rate." Short, free (public domain). [Free] [Intermediate]
• Marginal Revolution University: the Money & Monetary Policy videos (mru.org): free, short, well-made; the money multiplier, the tools, and the transmission mechanism at exactly this level. [Free] [Beginner]
• Zerodha Varsity: "Currency, Commodity, and Government Securities" and macro modules (India): for the RBI toolkit, the G-sec market, and the SLR/CRR plumbing in the Indian context. [Free] [Beginner]
• Bank of England, "Money creation in the modern economy" (Quarterly Bulletin, 2014): the central-bank paper that officially endorsed the endogenous-money view ("loans create deposits") over the multiplier; the authority behind §4.5–4.6. Free PDF. [Free] [Intermediate]
• Aswath Damodaran: his posts and classes on the risk-free rate and "the Fed and interest rates" (free on his NYU page and blog): the bridge from the policy rate to the discount rate, from the valuation master himself; the perfect complement to §4.10. [Free] [Intermediate]
• Odd Lots (Bloomberg podcast): for hearing practitioners argue about live policy in plain language; search episodes on QE/QT, the reserves-vs-money debate, and the transmission mechanism. [Free] [Intermediate]
• Going deeper: John B. Taylor, "Discretion versus Policy Rules in Practice" (Carnegie-Rochester Conference Series, 1993; free PDF from Stanford) [Free] [Advanced]: the original Taylor-rule paper behind Worked Example 4; watch a one-line rule shadow decades of Fed policy, and note exactly where central banks deliberately depart from it (which is the whole lesson).
• Going deeper: Perry Mehrling, The New Lombard Street: How the Fed Became the Dealer of Last Resort (Princeton, 2011) [Paid] [Advanced], the "money view" of central banking: the central bank not merely as rate-setter but as the market-maker and dealer of last resort whose backstop prices liquidity itself; the deepest complement to §4.1's LOLR and §4.5's balance-sheet mechanics.

The Modern Analyst's Addendum

Everything above teaches this skill from first principles, by hand, which 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": "The three hats and the lender-of-last-resort function, mandates and decision bodies, the corridor and OMOs, the CRR/SLR/guidance/FX toolkit, QE and the reserves-versus-broad-money distinction, the transmission mechanism and its long and variable lags, reading a statement, and the rate as gravity",
  "use": "The one task in this module that a language model does genuinely well is the one §4.8 says carries most of the information: comparing THIS statement to LAST statement. Paste both and ask for a sentence-level diff — what was deleted, what was added, what was reworded — and you get in seconds the thing a trained reader spends twenty minutes extracting. Two disciplines make that safe. First, get the diff mechanically, not by asking a model to remember the previous statement; supply both texts yourself, because a recalled prior statement is a fabricated prior statement and the diff is then worthless in a way you cannot see. Second, treat the model's TONE verdict as a hypothesis and the diff itself as the evidence: 'deleted: additional policy firming may be appropriate' is a fact you can point at; 'this is dovish' is a reading you own. Beyond the diff: (1) REACTION-FUNCTION QUESTIONING — 'given this statement, what data release would most change this committee's next decision, and in which direction' is the conditionality §4.8's press-conference section is really after. (2) VOCABULARY DRILLING — have it generate hawkish/dovish rewrites of the same sentence until the gradations are obvious. (3) MECHANISM CHECKS — 'walk the T-accounts for the RBI selling $5bn of reserves, then for the sterilising OMO' is §4.4 in bookkeeping form and it is checkable. What it must never supply: the current repo rate, the current fed funds range, the corridor's current width, the current CPI print, or the size of anyone's balance sheet. Those change, the model's answer will sound identical whether it is current or eighteen months stale, and a policy read built on a stale level is wrong in a way that survives every other check you run.",
  "tools": ["Chat assistants — Claude, ChatGPT for statement diffs (both texts supplied), reaction-function questioning and hawkish/dovish drilling", "Document/PDF AI — for MPC minutes, the RBI Monetary Policy Report, FOMC minutes and the SEP's projection tables", "Coding assistants — for the Taylor-residual series, the statement-diff pipeline and the implied-path comparison below"],
  "prompt": "Here are two central-bank statements, PRIOR and CURRENT, pasted in full. Produce a sentence-level diff: every deletion, every addition, every reworded clause, quoted exactly, with no paraphrase. Then, separately and clearly labelled as interpretation, say which direction each change tilts and what data release would most plausibly reverse it. Do not tell me the policy rate, the inflation rate, or any other current level — I have pulled those from the source myself and they are not your job.",
  "verify": "Three checks, in order of how often they catch something. (1) The diff must be quotable. Every claimed deletion or addition must appear verbatim in one text and not the other — search for the string; a paraphrased 'the Committee softened its language' is not a finding and cannot be verified. (2) Every level in your final read must carry a source and a date: the repo rate from the RBI's press release with that release's date, the fed funds range from the FOMC statement with its date. §4.3 and §4.8 both quote levels marked 'as of mid-2026 — verify' for exactly this reason, and the module's own instruction is the standard. (3) Arithmetic on those levels — the corridor width, the real policy rate, the Taylor residual, the number of cuts a market path implies — you compute yourself, because these are three-line calculations and a wrong one is invisible inside fluent prose. Then the deeper verification, which is about you rather than the tool: §4.7's long and variable lags mean no read of a statement can be scored against the next three months of data. A dovish read that is followed by falling yields was not thereby confirmed. Judge a read on whether it named the right conditionality, not on whether the next print went your way.",
  "diy": "The gate is unaided: you state the corridor from a policy rate, convert a nominal rate to a real one by both the approximation and exact Fisher, compute a Taylor prescription and the residual against the actual rate, price a bond at two yields and get the percentage move, count the cuts a dot path and a market path each imply, and classify a statement change as hawkish or dovish with a reason. Nothing in this callout is admissible in that room."
}

Modern Data Analysis

By hand first. You traced a 50-bp cut from the corridor to a household's EMI and then to the lending bank's margin; you did the QE T-accounts for a bank seller and a non-bank seller and got different broad-money answers; you computed a Taylor prescription and a real repo rate and concluded policy was easier than both benchmarks; you repriced a 10-year note for a 100-bp move and watched duration overstate the loss. Keep all of it, and the T-accounts most of all. The reserves-versus-broad-money confusion of §4.5 is not curable by argument, only by writing both sides of three balance sheets.

Today's workflow. Three changes, and the first is the one that turns §4.8 from an art into a procedure. The statement read becomes a diff. What a trained reader extracts from a policy statement is almost entirely the delta against the previous one: a deleted clause, an added conditional, an adjective moved. Held as two text files and differenced sentence by sentence, that delta is produced mechanically and completely, instead of by re-reading and hoping you noticed. The Fed publishes every statement, minute and SEP at stable URLs on federalreserve.gov and the RBI publishes every MPC resolution, Governor's statement and minutes as press releases on rbi.org.in, so the archive is buildable once and appended to on a known calendar. Second, "hawkish" and "dovish" become a number. §4.8 teaches the vocabulary; the Taylor residual, the actual policy rate minus what a stated rule prescribes from stated inputs, turns it into a signed quantity you can plot, and a plot of that residual through a cycle says more than any run of adjectives. Third, the policy path becomes two curves rather than one. §4.8's professional move is comparing the committee's own projection to the market's; done properly that is the SEP median dot path against the fed funds futures / OIS-implied path, and for India the RBI's projections against the INR OIS curve. The gap is the object of interest, and it is a subtraction.

Tools & sources (IN + US). pandas for the series, difflib for the statement diff, numpy and matplotlib for the residual; India: the RBI Monetary Policy Committee resolution and the Governor's statement on decision day, the MPC minutes (published about fourteen days later, with each member's vote and rationale, the split §4.2 says is information), the published bi-monthly MPC calendar so the diff pipeline knows when to run, the Monetary Policy Report (bi-annual, carrying the RBI's own CPI and growth projection paths), RBI DBIE for the repo/SDF/MSF history and the Weekly Statistical Supplement for the RBI's balance sheet and liquidity position, the RBI's daily money-market operations releases for the WACR and the LAF, CCIL for the call and TREPS money-market data and the G-sec curve, and FBIL for the overnight MIBOR and the INR OIS benchmark, since the OIS curve is the Indian analogue of fed funds futures and is where the market's implied policy path comes from; MOSPI for the CPI print the target is defined on; US: the FOMC statement, minutes and the quarterly SEP (the dot plot is published as a table, not only as a picture), the FOMC's published meeting calendar, FRED by id for FEDFUNDS and DFF (the effective rate), SOFR and IORB (the market rate and the floor of §4.3), WALCL (the balance sheet whose expansion is QE, and whose run-off is QT), BOGMBASE and M2SL (the base-versus-broad pair that makes §4.6's multiplier computable), PCEPILFE (the index the 2% target is actually defined on, not CPI), UNRATE, GDPC1 and GDPPOT (real output against CBO potential, for an output-gap proxy), and DRTSCILM (the SLOOS net percentage of banks tightening commercial-and-industrial lending standards, §4.7's credit-availability channel, measured). For the neutral rate, which is the whole fragility below: the New York Fed publishes the Laubach-Williams and Holston-Laubach-Williams r* estimates as a free, dated, revised series with a documented model, so a US Taylor residual can be built on a published r* whose vintages you can inspect. India has no continuously-updated official equivalent; the RBI has published natural-rate estimates in Monetary Policy Report boxes and RBI Bulletin articles, and you must cite the specific box and its date. That asymmetry is the lesson, not a footnote. Everything US above is free and machine-readable. On the India side the RBI, MOSPI, CCIL and FBIL publications are free, but FBIL's OIS benchmarks arrive as daily files rather than a queryable history, so a long OIS-implied-path series is either assembled by you day by day from the day it occurs to you, or bought: CMIE, CEIC, Bloomberg or Refinitiv. You cannot backfill it. Start the archive the week you read this.

# Taylor prescription, the residual, and what the residual is actually sensitive to — R3
def taylor(rstar, pi, pistar, gap, a=0.5, b=0.5):
    return rstar + pi + a * (pi - pistar) + b * gap
def residual(actual, *args, **kw):          # >0 = tighter than the rule, <0 = easier
    return actual - taylor(*args, **kw)
print(f"{residual(5.25, 1.5, 4.5, 4.0, 0.5):+.2f}",     # -1.25  WE4, India
      f"{residual(6.50, 2.0, 6.0, 4.0, 0.0):+.2f}",     # -2.50  P14
      f"{residual(6.00, 1.5, 5.0, 4.0, 1.0):+.2f}")     # -1.50  P18
b0 = taylor(1.5, 4.5, 4.0, 0.5)                          # the partials are exact
for lbl, alt in (("r*", taylor(2.5, 4.5, 4.0, 0.5)), ("pi", taylor(1.5, 5.5, 4.0, 0.5)),
                 ("gap", taylor(1.5, 4.5, 4.0, 1.5)), ("pi*", taylor(1.5, 4.5, 5.0, 0.5))):
    print(lbl, f"{alt - b0:+.2f}")           # +1.00   +1.50   +0.50   -0.50
span = [residual(5.25, rs, 4.5, 4.0, g) for rs in (0.5, 2.5) for g in (-1.0, 1.0)]
print(f"{min(span):+.2f} to {max(span):+.2f}")           # -2.50 to +0.50 — both signs

Verify. Prove the engine on the module's own worked figures first. Worked Example 4: real repo by approximation 5.25 − 4.5 = +0.75%, exact Fisher 1.0525 / 1.045 − 1 = +0.7177%, Taylor prescription 1.5 + 4.5 + 0.5(0.5) + 0.5(0.5) = 6.50%, residual −125 bp. Problem 14: prescription 2 + 6 + 1 + 0 = 9.00%, residual against a 6.50% repo −250 bp. Problem 18: exact real repo 1.06 / 1.05 − 1 = +0.9524%, prescription 7.50%, residual −150 bp. Worked Example 1: EMI at 9.00% ₹44,986.30, at 8.50% ₹43,391.16, monthly saving ₹1,595.14, annual ₹19,142, with the compounding factors 1.0075²⁴⁰ = 6.0092 and (1 + 0.085/12)²⁴⁰ = 5.4412. Worked Example 5: price at 5% $922.78 (coupons 308.87 plus principal 613.91), change −7.72%, Macaulay duration 8.435, modified duration 8.111, linear estimate −8.11%, convexity error +0.39 pp. Problem 15: $955.48, −4.45%, modified duration 4.580. Problem 16: new NIM 4.40%. Problem 4: multiplier 264 / 48 = 5.50×. Problem 12: −2.8% and −57.0%. Problem 17: −20.0% and −11.1%. Every one reproduced here. Then verify what the engine gives you that the module leaves as a sentence: Worked Example 1(c) states the shares (60% of the loan book repricing down 50 bp against 15% of deposits) but not the margin number they imply, and −50 bp × (0.60 − 0.15) = −22.5 bp is that number. The general form is worth memorising because it holds exactly: ΔNIM = Δrate × (share of assets repricing − share of liabilities repricing), which is Problem 16's +40 bp from 100 bp × (0.70 − 0.30) and Worked Example 1's −22.5 bp from the same one-line rule.

Quantitative lens

§4.8 gives you "hawkish" and "dovish" and a vocabulary for applying them. The Taylor residual gives you the same judgement as a signed number, and the number is more useful because it can be plotted, differenced, and argued with. All figures below come from the engine above run on the module's own stated inputs; they are properties of that rule and those inputs, not measurements of any central bank's intent.

Say it without adjectives. Define the residual as actual policy rate − Taylor prescription, so a positive number means policy is tighter than the mechanical benchmark and a negative one means easier. Run it on the module's three parameterisations. Worked Example 4's India setting: prescription 6.50%, actual 5.25%, residual −125 bp. Problem 14: prescription 9.00%, actual 6.50%, residual −250 bp. Problem 18: prescription 7.50%, actual 6.00%, residual −150 bp. Three cases, three numbers, and each one carries what a paragraph of adjectives was trying to carry, with the added benefit that "policy is 125 bp easier than the rule" invites the reply "which rule, and with what r*", whereas "the RBI is dovish" invites nothing.

Why the rule has a coefficient at all: the Taylor principle, demonstrated. The rule's 0.5 × (π − π*) term looks like a tuning knob. It is the entire point of the rule, and the arithmetic shows why. Differentiate the prescription: ∂i/∂r* = +1.00, ∂i/∂π = +1.50, ∂i/∂(output gap) = +0.50, ∂i/∂π* = −0.50, all exact. The +1.50 is the one that matters. Because the prescribed nominal rate rises 1.5 points for every 1 point of inflation, the prescribed real rate rises 0.50 points. Set the coefficient to zero instead and the prescription becomes r* + π: the nominal rate still rises one-for-one with inflation, but the real rate does not move at all, +0.00 pp, so policy leans against nothing and inflation is free to run. At a coefficient of 1.0 the real rate rises a full +1.00 pp. That is the Taylor principle in three lines: a rule that does not raise the real rate when inflation rises is not a stabilising rule, whatever its nominal response looks like. It is also why §4.7's transmission chain matters — the real rate is what Step 4 responds to.

And now the limit, which is large enough that it must be stated before the number is used. The residual depends on two quantities nobody observes: the neutral real rate r* and the output gap. Hold the module's inflation figures and let r* range over 0.5% to 2.5% and the output gap over −1% to +1% (both entirely defensible ranges for the India case) and the residual spans −2.50 pp to +0.50 pp, a 300 bp window that contains zero and both signs. At r* = 0.5% with a −1% gap the same 5.25% repo is tighter than the rule; at r* = 2.5% with a +1% gap it is 250 bp easier. Put the question the other way and it is sharper still: the neutral real rate that would make a 5.25% repo exactly rule-consistent on the module's other inputs is 0.25%. So "policy is 125 bp too loose" is, quantitatively, at least as much a statement about your r* estimate as about the RBI, which is exactly why the module says read it, don't obey it, and why the New York Fed publishing HLW as a dated, revisable series with its model attached is worth more than any single point estimate.

There is a repair, and it is algebra rather than judgement. The level of the residual is contaminated by r*; its change is not. Difference the residual between two dates and, provided you hold r* and the target π* fixed, the r* term cancels identically: Δresidual = Δ(policy rate) − 1.5 × Δπ − 0.5 × Δ(output gap). Whatever you believe r* to be, it drops out. So the honest way to use this tool is to say "the residual moved 60 bp tighter over the last two quarters" rather than "policy is 125 bp loose". The first is robust to the thing you do not know, the second is built on it. This is the same instinct §4.8 applies to statements (read the change, not the level) arriving through the arithmetic.

One more number the module leaves implicit: transmission is a weighted average, not a rate cut. §4.7 and Problem 7 establish that the same 50-bp cut reaches an EBLR-linked borrower in full, an MCLR borrower as about 15 bp after two quarters, and a fixed-rate borrower not at all. Weight those by book composition and the effective pass-through falls out: a book that is 60% EBLR, 30% MCLR and 10% fixed transmits 34.5 bp of a 50-bp cut, 69%; at 40/40/20 it transmits 26.0 bp, 52%; at 80/15/5 it transmits 42.2 bp, 84%. The headline "the RBI cut 50 bps" is therefore never the number that reaches the economy, and the gap between the announced cut and the delivered one is a bank-by-bank, book-by-book fact that the RBI's 2019 EBLR mandate was designed to shrink.

Honest limits. The Taylor rule is a benchmark and not a description of any committee's decision process: no central bank has ever claimed to follow it, the RBI operates a flexible framework that explicitly weighs growth, the rupee and financial stability, and a residual is therefore a comparison to a yardstick rather than a measurement of a deviation from duty. The coefficients 0.5 and 0.5 are Taylor's original choices, not estimates; changing them changes every number above, and a rule fitted to a country's own history is a different object with different properties. The output-gap input is doubly unobservable, since potential output is itself a model output that is revised for years afterwards, so today's residual will be restated. The figures here use the module's stated inputs, which are marked as illustrative and as-of, and none of them is a claim about where any policy rate is now; pull the current repo rate from the RBI's press release and the current target range from the FOMC statement, both with their dates, before you compute anything. And the deepest limit is §4.7's: even a perfectly measured residual tells you the stance, not the effect, because the effect arrives twelve to twenty-four months later by an amount that is not knowable in advance. A residual that says policy is easy today says nothing verifiable about output next quarter, and an analyst who treats it as a forecast has re-imported precisely the over-confidence the lags forbid.

Do it in code: archive both central banks' statements from their published calendars and diff each against its predecessor; build the residual as a series and read its change, not its level; put the committee's own projected path and the market-implied path on one chart because the gap is the object; and compute effective pass-through from book composition rather than quoting the headline cut. See G1/MS1.03, G1/QM1.02 and G2/DA1.05.

Where this goes next: galaxy cross-links

• G1 · MS1.03 Central Bank Operations & Repo: the corridor, the standing facilities and the repo mechanics of §4.3 at operating-desk depth, including how a liquidity operation moves the overnight rate without touching the headline rate.
• G1 · MS1.02 Payment System Plumbing: why reserves are the settlement asset and what actually happens between two banks at end of day, which is the concrete content of "banker to the banks."
• G1 · QM1.02 Hypothesis Testing & Regression: estimating a reaction function from history instead of assuming Taylor's coefficients, and the standard errors that decide whether an estimated coefficient supports the conclusion you want to draw from it.
• G2 · DA1.05 Statistics in Code: differencing, rolling windows and the discipline of testing a claim about a series against the series rather than against a chart of it.
• G3 · AI0.06 Failure Modes & Guardrails: the stale-level failure named above, why it is invisible to every downstream check, and how to build a workflow in which a model is never the source of a number.

Teach it back & journal

Feynman prompt. Write a one-page explainer for a smart 15-year-old titled "Who decides interest rates, and why should you care?" You must convey, without jargon: that a central bank sets only one rate (the price of overnight money between banks) and that everything else (your family's home-loan EMI, the return on a fixed deposit) follows from it through a long chain; that changes take a year or two to fully work (so the people setting rates are always steering with a delay); and the one idea that ties it to investing: when this rate goes up, everything worth owning (houses, stocks, bonds) tends to get cheaper, because a higher rate is like stronger gravity on prices. If your page needs the words "transmission mechanism," "quantitative easing," or "duration," you don't own it yet — say it in plain words.

Journal prompt. Three entries. (1) Pull the current repo rate and the current fed funds range (rbi.org.in; federalreserve.gov) and write today's date beside each. Then, in one line each, note the direction of the last move and what the stance/guidance says about the next one. This is the first entry of a habit you'll keep for years; the numbers you write today are the "as of" you'll verify against later. (2) Take one company you've studied (from the starter bank or your own list) and write two sentences: how would a 200-bp rise in its economy's policy rate affect (i) its cost of capital / valuation, and (ii) its actual business (borrowing costs, customer demand)? Notice whether the valuation effect or the business effect is larger. That judgment is the heart of macro-aware analysis. (3) Reflect honestly: before this module, did you believe "the central bank prints money and that causes inflation"? Write down what you now think is right and wrong about that sentence, in your own words. This is the misconception the module most wanted to fix, and writing the correction is how it sticks.

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