India Macro + Dashboard Capstone

Phase 7 · Week 63 · ~20 hours · Phase capstone (Milestone M8)

India buys roughly 85–88% of its crude oil from abroad. So when oil rises $10 a barrel and stays there, three things move at once: the annual import bill widens by about $13–15 billion, fuel prices push up the CPI the RBI targets, and the government has to choose between subsidising the pump and letting consumption take the hit. One price, three parts of the economy.

Where it sits. The final module of the macro phase, and its synthesis. The six modules before it built the machine in the abstract and for the world: M7.01 the economic machine (spending = money + credit; the debt cycle; S = I); M7.02 central banks and the toolkit (repo, the corridor, CRR/SLR, transmission, QE); M7.03 inflation, rates and the yield curve (CPI/core/PCE/WPI, the curve as a signal, rates as gravity); M7.04 fiscal policy and sovereign debt (the r-vs-g debt dynamics, "denomination is destiny"); M7.05 cycles, bubbles and crises (Minsky, the growth–inflation 2×2, leading/coincident/lagging indicators); M7.06 FX, flows and the dollar system (the impossible trinity, the balance of payments, EM sudden stops, commodities as macro signals). Two things happen to all of that here. First, the general machinery turns on one specific economy, India, because you cannot analyse an Indian business without a working model of the economy it lives in, and India's macro has enough structural peculiarities (a managed rupee, a captive-buyer bond market, a demographic bulge, a domestic-demand engine, a fiscal picture that looks alarming until you see it is almost all in rupees) that "generic emerging market" is a wrong model. Second, you get the standing instrument every macro-literate analyst runs for the rest of their career: a ~18-indicator dashboard, read not as a wall of numbers but as rate-of-change and versus-expectations, mapped onto the debt-cycle stage and the growth–inflation quadrant, and compressed into a one-page regime memo that ends in a portfolio stance. The two capstone deliverables (build the dashboard from primary sources, and write the regime memo) are the M8 milestone and the certification of competency C8. By the end you can read the weather, for the world and for India, and say what it means for the price of everything.

Suggested hour budget. Core content (the India synthesis + the dashboard-and-regime method): 8h · worked examples (the oil→twin-deficit channel, reserves adequacy, reading a regime): 3h · practice set: 4h · the capstone mini-project (assemble the live dashboard + write the regime memo): 3h · flashcards + mastery check: 2h. The capstone is deliberately light on new mechanics and heavy on assembly. This is where the phase's parts click together.

App link. This module completes the Macro & Cycles branch (competency C8) and powers two things in the app. The scenario-event engine turns the India-flow mechanics into something playable. The "oil shock," "rupee sell-off," "FPI sudden stop," "rate cut," and "fiscal-slippage" events that fire mid-simulation each ask what happens to the CAD, the rupee, inflation, the fiscal, and your portfolio, and what do you do? The dashboard widget + regime-classifier drill operationalises the method: a snapshot of the ~18 indicators comes in (as levels, deltas, and surprises), and you must name the growth–inflation quadrant, place the debt-cycle stage, and pick the defensible stance, auto-scored against the regime logic in §4.9–4.10. The regime memo you write here is the template the app's rubric grades every later macro-overlay memo against. This is where the app connects the macro weather to a single business's discount rate and demand, closing the loop the whole program is built on.

Learning objectives

By the end you can:

1. State the India macro thesis as one coherent frame: a young, domestic-demand-led, services-heavy, rapidly financialising economy with a real growth engine and three honest constraints (jobs, manufacturing share, human capital). Hold the bull and the caveat in the same sentence, without collapsing into either boosterism or declinism.
2. Explain the RBI's flexible-inflation-targeting (FIT) framework in operation: the repo rate as the anchor, the SDF/MSF liquidity corridor, and the CRR and (India-specific) SLR. Read a live MPC resolution for its India-specific signals: the stance, the vote split, the headline-CPI framing, and the liquidity stance underneath the headline rate (building directly on M7.02).
3. Explain why India runs a managed float rather than a fixed or a freely-floating rupee, describe how RBI FX intervention works and its domestic-liquidity side effect, and compute and interpret the reserves-adequacy metrics: months of import cover, reserves-to-short-term-external-debt (Guidotti–Greenspan), and the IMF ARA composite.
4. Lay out India's growth engine and its limits: the demographic dividend (and the jobs caveat: it is a dividend only if employed), the "India Stack" (GST formalisation, the IBC's credit discipline, UPI/digital public infrastructure, Aadhaar), domestic-demand-led growth (its resilience and its ceiling versus the East-Asian export model), and the SIP/financialisation-of-savings flow that now lets domestic money cushion foreign outflows. Set those alongside the constraints (labour-force participation, the stuck ~17% manufacturing share, and human-capital/learning gaps).
5. Read India's fiscal picture like an analyst, meaning the deficit trajectory and the debt level, and explain the single fact that makes it resilient rather than fragile: it is overwhelmingly rupee-denominated and domestically held, so India cannot suffer the dollar-debt sovereign crisis that felled Argentina or Sri Lanka (the "denomination is destiny" point from M7.04), while still naming the real constraint (the interest-to-revenue burden).
6. Trace India in global flows: the structural current-account deficit and why it is structural; the oil-import sensitivity and the twin-deficit-via-oil channel (quantified with a rule of thumb); FPI flows and the significance of the JPMorgan/Bloomberg bond-index inclusion; and the resilience buffers built since the 2013 taper tantrum that moved India off the "Fragile Five" list.
7. Navigate the primary Indian data sources precisely: RBI DBIE, MOSPI/NSO (CPI, IIP, GDP, PLFS), the Ministry of Finance Economic Survey and Union Budget, CMIE, and PRS Legislative Research, knowing for each what it releases, when, and exactly where to click.
8. Build and read the ~18-indicator macro dashboard as rate-of-change and versus-expectations, map the reading onto the growth–inflation 2×2 and the debt-cycle stage, and write a regime memo that names the regime, places the cycle, and ends in a defensible portfolio stance with its falsification triggers.

Prerequisites & connections

Builds on. The whole of Phase 7. M7.01: the debt cycle and S = I are the frame the dashboard maps onto and the reason the current account (S − I) matters. M7.02: the entire RBI toolkit section (§4.2) is a direct application. The repo, the SDF/MSF corridor, CRR/SLR, and the managed-float FX-intervention mechanics are taught there and operated on India here; you cannot read §4.2 without it. M7.03: CPI vs core vs WPI, the yield curve, and "rates are gravity" are the inflation/policy and financial-conditions blocks of the dashboard. M7.04: the r-vs-g debt-dynamics arithmetic and "denomination is destiny" are the whole of the fiscal section (§4.6); the resilience claim is that argument applied to India. M7.05: Minsky's stages, the leading/coincident/lagging taxonomy, and the growth–inflation 2×2 are the regime map (§4.9); "assets are priced off surprises" is why the dashboard reads versus-expectations. M7.06: the impossible trinity, the balance of payments, EM sudden stops, and commodities-as-signals are the global-flows section (§4.7); the rupee section (§4.3) is the impossible trinity's Indian answer. From earlier phases: M3.02–M3.03 (the equity risk premium, the country-risk premium for India, and the cost of capital), because macro is where the discount rate's inputs come from; M0.03 (real vs nominal, Fisher), because India's higher inflation makes the real-vs-nominal distinction matter more than it does for a US investor; and M5.01–M5.02 (banks and NBFCs), because CRR/SLR, CASA, credit growth, and ALM are the plumbing through which all of this reaches a lender's P&L.

Feeds forward. M8 (the entire rapid-analysis system): every teardown and deep dive from Phase 8 on carries a macro-overlay line, asking where we are in the cycle, what the discount-rate environment is, and what macro variable could break the thesis. This is where you learn to write that line, and the dashboard is the standing input to it. Phase 9 (portfolio and risk): the regime read is a top-down input to position sizing and to the "when to raise cash" question, and the scenario events in the simulator are those shocks. Competency C8 ("understand the macro machine") is certified here: the milestone is the dashboard built from primary sources plus the mechanism exam plus the regime memo graded against the rubric. The macro-dashboard artifact (artifacts/macro-dashboard.md) is the reusable, per-indicator reference you will run monthly for years, produced alongside this module; §4.9 teaches the method, the artifact is the manual.

The one-sentence version. India is a young, domestic-demand-led, fast-financialising economy whose central bank targets 4% CPI with a managed rupee and a deep reserve buffer, whose scary-looking fiscal deficit is defused by being almost entirely in rupees, and whose main external vulnerability runs through oil, and the way you keep this and every other economy in view is a small dashboard of indicators read as change-and-surprise, compressed into a one-page regime call that tells you what the weather implies for the price of risk.

Core teaching content

4.1 The India macro thesis, in one frame

Before any indicator, you need a model of the economy: a compressed thesis you can state in four sentences and then spend the rest of the material qualifying. Here it is, and everything below is its expansion:

India is the world's most populous country and its fastest-growing large economy, powered by a young population and domestic demand rather than exports, with a distinctively services-led growth path and a fast financialisation of household savings that is deepening its capital markets. Its central bank runs a credible 4%-CPI inflation target and a managed float backed by a large FX-reserve buffer. Its fiscal deficit and public debt look high by emerging-market standards but are overwhelmingly rupee-denominated and domestically financed, which removes the classic sudden-death sovereign-crisis risk. Its growth is real but not yet inclusive: the binding questions are whether it can generate enough good jobs for its young workers, lift its manufacturing share, and close its human-capital gap fast enough to convert a demographic bulge into a demographic dividend.

Every clause there is a claim an analyst must be able to defend and to attack. The discipline that matters is holding the bull case and the honest caveat in the same breath. The lazy versions are everywhere: the boosterish "India is the next China, buy everything" and the declinist "jobless growth, it will never industrialise, it's all hype." Both are analytically useless, because both are unfalsifiable moods. The analyst's version is structural: here is the growth engine, here is exactly what it depends on, here is what would tell me it is working or failing, and here is where the price already reflects the optimism.

Note the last clause. India is a famously expensive market (its large-cap index has traded at a premium P/E to most emerging and even developed markets for years, a "quality/growth premium" the market assigns to its consistency and its domestic-flow support), so even a correct bull thesis can be a poor investment at the wrong price. That tension between a great economy and a demanding valuation is the standing backdrop to every Indian teardown you will do, and it is why macro literacy and valuation discipline (Phase 3) have to travel together.

Three structural facts frame everything below, and they distinguish India from the generic EM template:

• It is domestic-demand-led, not export-led. Private consumption is roughly 57–60% of GDP (illustrative, ~FY24/25 scale; check at MOSPI); goods-and-services exports are around 21–22% of GDP, far below the 30–50%+ that characterised the East-Asian export miracles at comparable stages. This is simultaneously India's great shock-absorber (a global trade recession hurts it far less than it hurts Korea, Vietnam or Germany) and its great speed limit (it has not, so far, built the manufacturing-export jobs machine that pulled hundreds of millions out of farms in China).
• It grew up services-first. India's comparative advantage emerged in IT services, business-process outsourcing, and now global capability centres, which are high-productivity, English-language, tradeable services, rather than in low-cost manufacturing. Services are ~54% of GVA and the bulk of exports' dynamism. This is a genuinely unusual development path (most economies industrialise before they "servicify"), with real strengths and one central weakness: high-end services employ the educated few, not the unskilled many coming off the farms.
• It is financialising fast. For decades Indian households held their savings in gold and real estate. That is now shifting, structurally, into financial assets: bank deposits, and increasingly equities via mutual funds. The monthly SIP (Systematic Investment Plan) flow into equity funds is a standing order that buys the market every month regardless of the news, and it has grown into a large, sticky, domestic bid (§4.4). This single fact has changed how the Indian market behaves in a foreign-selling episode, and it is one of the most important structural shifts of the last decade.

Hold this frame. Now we build it out, piece by piece, and the pieces are exactly the dashboard's India add-ons: the RBI and rates, the rupee and reserves, growth and its constraints, the fiscal, and the external flows.

4.5 The constraints, honestly

A world-class analyst is defined partly by refusing to launder away the hard parts. Three constraints could stall the thesis, and you must be able to state each with its evidence — not as declinist mood, but as structural risk.

(1) Jobs and labour-force participation. The headline unemployment rate (PLFS, ~3–4%) looks fine and is misleading, for three reasons. First, in a poor country with thin social safety nets, most people cannot afford to be openly unemployed. They take any work, so distress shows up as underemployment and informality, not unemployment. An estimated ~80–90% of India's workforce is informal, much of it low-productivity self-employment and casual labour. Second, labour-force participation is low, and female LFPR especially so, long stuck around 20–25% before a recent (partly definitional, partly rural-distress-driven) rise; a large share of working-age women are simply not in the labour force, an enormous unused growth reserve. Third, agriculture still employs ~42–45% of the workforce while producing only ~15–18% of GDP, a vast pool of low-productivity labour that development is supposed to move into higher-productivity factory and service jobs. That transition has been slow. The "jobless growth" debate, over whether India's fast GDP growth is generating enough good jobs, is genuine and unresolved, and it is the direct threat to converting the demographic bulge into a dividend (§4.4).

(2) The stuck manufacturing share. Manufacturing has been stuck at roughly 15–17% of GDP for decades, despite the explicit "Make in India" goal of lifting it to 25% (a target repeatedly set and missed). This matters because manufacturing is historically the great absorber of unskilled labour off the farm, the escalator every industrialised economy rode. India's services-first path skipped that escalator, a pattern economists call "premature deindustrialisation." The policy response is the PLI (Production-Linked Incentive) schemes, subsidies tied to incremental output in targeted sectors (electronics, pharma, autos, solar), which have had visible early wins (notably smartphone assembly and exports). But turning assembly into deep manufacturing, and lifting the aggregate share, is a decade-long structural project against real headwinds (land, labour laws, logistics costs, and a global environment less open to trade than the one China rode). The analyst watches the manufacturing GVA share and the PLI outcomes as a multi-year tell on whether the jobs problem is being solved.

(3) Human capital. A demographic dividend requires not just bodies but skilled, healthy bodies, and here the evidence is sobering. India's learning outcomes lag badly: the annual ASER survey has for years shown that a large share of rural Grade-5 children cannot read a Grade-2 text or do simple division. That is a learning deficit, not merely a schooling one. Malnutrition and stunting remain high enough to impair lifetime productivity. Higher education is uneven, and the skilling gap between what the education system produces and what employers need is wide. Human capital is the slowest-moving variable of all, because you cannot fix a cohort's early-childhood learning retroactively, which is why it is simultaneously the most important long-run driver and the least amenable to a quick policy win. It is the deepest of the three constraints.

The honest synthesis: India's growth is real, but its inclusiveness and its durability depend on solving jobs, manufacturing, and human capital. None of the three is guaranteed, and all of them move slowly. The bull thesis is not wrong; it is conditional, and naming the conditions precisely is the analyst's job. (There is also a live distributional debate, the "K-shaped" recovery argument that the formal, affluent, urban top has pulled away from the informal, rural, mass bottom. It shows up in the divergence between premium-consumption strength and mass-consumption softness, and you can read it directly in company results across the consumer sector, M5.05.)

4.8 The data sources: exact navigation

The whole program's rule 7 is that primary sources outrank every summary. For India macro, that means knowing the handful of primary sources cold: what each releases, when, and where to click. This is the toolkit you will use to build the dashboard.

• RBI DBIE, the Database on the Indian Economy (dbie.rbi.org.in, being migrated into the RBI Data Warehouse / "Sarthi" / next-gen portal; check the current URL). The RBI's own statistical warehouse and your first stop for anything monetary or external: FX reserves and money supply (from the Weekly Statistical Supplement, WSS, released each Friday), M0/M3, bank credit and deposit growth, the corridor rates, the balance of payments and CAD (quarterly), and the long-run Handbook of Statistics on the Indian Economy (annual). The bi-monthly MPC resolution, Governor's statement, and Monetary Policy Report live on the main RBI site under Press Releases / Monetary Policy. This is where reserves, credit growth, money, and the corridor come from.
• MOSPI / NSO, the Ministry of Statistics & Programme Implementation / National Statistical Office (mospi.gov.in). India's official statistics agency and the source for the real-economy core:
• CPI (Consumer Price Index, the RBI's target variable), released ~12th of each month for the prior month.
• IIP (Index of Industrial Production, the monthly factory-output proxy), released ~last working day of the month, for the month ~6 weeks prior (it lags).
• GDP / National Accounts: quarterly (released end-Feb, May, Aug, Nov) plus annual estimates; this is where real GDP growth comes from.
• PLFS (Periodic Labour Force Survey): the official unemployment rate, LFPR, and worker-population ratio (annual, with quarterly urban rounds). This is where growth, inflation, industrial activity, and jobs come from.
• Ministry of Finance: the Economic Survey and the Union Budget (indiabudget.gov.in, and the Department of Economic Affairs). Two flagship annual documents, released together at the end of January / start of February:
• The Economic Survey: the government's flagship economic analysis of the year just gone and the outlook (released the day before the Budget). The single best official narrative of the Indian economy; read its overview chapter each year.
• The Union Budget, the fiscal numbers: the deficit, borrowing, receipts and expenditure, delivered Feb 1. This is where the fiscal picture (§4.6) comes from.
• CMIE, the Centre for Monitoring Indian Economy (cmie.com; a paid database, sold as its Economic Outlook and Prowess products). CMIE's most-cited public output is its higher-frequency unemployment rate (from the Consumer Pyramids Household Survey, CPHS), a monthly, faster read than the official PLFS, widely reported in the press even if you don't subscribe. Also strong for company-level and investment-project (capex-pipeline) data. Mark it [Paid] but know its headline unemployment number, because the market watches it.
• PRS Legislative Research (prsindia.org; free, non-partisan). The best plain-English decoder of the Budget and of legislation: its annual "Union Budget Analysis," state-finances briefs, and bill trackers turn the raw fiscal documents into something readable, with the historical context and the numbers put in proportion. When the Budget lands, read the PRS analysis alongside the primary document. [Free]
• The supporting cast (know they exist): the Office of the Economic Adviser (WPI, the Wholesale Price Index, M7.03); the Ministry of Commerce (monthly merchandise-trade data, the import/export bill); NSDL / CDSL and SEBI (FPI flow data); the GST portal / PIB press releases (monthly GST collections); and AMFI (mutual-fund and SIP-flow data, for the financialisation read of §4.4).

A discipline worth building now: for each dashboard indicator, know its primary source and its release calendar, so you are reading the number the day it comes out, in context, rather than a stale second-hand mention weeks later. The habit of pulling and plotting the series yourself, as M7.02's resources urged for FRED, is worth more than any summary.

4.10 The regime memo: from dashboard to stance

The dashboard is the input; the regime memo is the output: a one-page discipline that forces the read into a decision. It is the deliverable that certifies competency C8, and its structure is a fixed four-step chain you will reuse for the rest of your career:

Step 1. Read the indicators (change + surprise). Go group by group through the dashboard and write, for each, the rate-of-change and the surprise, not the level. "Growth: PMIs rolling over from 55 to 51, below the 52 consensus; retail sales decelerating. Inflation: CPI 3.2% and falling faster than expected. Policy: RBI cut 25 bps, dovish stance-shift. Financial conditions: curve steepening, spreads stable. Global: oil soft, rupee stable, VIX low." The output of Step 1 is a paragraph of deltas and surprises, not a table of levels.

Step 2. Name the regime (the 2×2 quadrant). Resolve the read into one of the four quadrants: is growth accelerating or decelerating, and is inflation accelerating or decelerating? Force a call: "disinflationary slowdown" or "reflationary recovery" or "stagflationary." Name it in two words. If the indicators conflict (they often will, and that is real), say which group you are weighting and why (leading indicators over lagging; surprises over levels).

Step 3. Place the debt-cycle stage. Where in the credit cycle does this sit? Early, mid, late, or recession/deleveraging? Cite the cycle-specific tells: the curve shape, spread direction, credit growth, and the central bank's posture (easing into a trough is early-cycle; hiking into an inversion is late-cycle). This is what distinguishes a Goldilocks-early ("buy") from a Goldilocks-late ("careful") read of the same quadrant.

Step 4. Draw the portfolio-stance implication (with triggers). Convert the regime-plus-stage into a stance, a direction of tilt rather than a trade list: risk-on or risk-off; long-duration or short-duration equities; add or cut cyclicals/credit; raise or deploy cash; and, for India specifically, the rupee/oil-sensitivity and rate-sensitivity implications. End with the falsification triggers: the specific indicators whose change would flip the regime call: "this read breaks if oil spikes through $90, if the curve re-inverts, or if the next two CPI prints surprise high." A regime call without its triggers is a mood; a regime call with triggers is a falsifiable thesis you can update (the Bayesian discipline of M6.03).

Two rules keep the memo honest. First, the memo is a range and a tilt, never a point forecast. You are locating the economy in a regime and leaning the portfolio, not predicting next quarter's GDP to a decimal; macro is largely unforecastable (M7.06's humility), and the memo's job is to be robust, not clairvoyant. Second, process over outcome (rule 8): you grade a regime memo on whether the read was defensible from the data and the logic was sound, not on whether the market went your way next month. A good regime call can be wrong for a quarter, and a lucky one can be right for the wrong reasons.

The full worked example, a filled model-answer regime memo on an illustrative dashboard snapshot, is the capstone in the Applied mini-project below. Study its structure, not its numbers; the numbers are frozen for teaching, the structure is what you keep.

Common mistakes & how experts think differently

Mistake 1: "India's debt-to-GDP is ~83%; it's heading for a Sri-Lanka-style crisis." The most common analytical error a generic-EM lens produces. The novice reads the ratio; the expert reads the denomination. India's public debt is ~95%+ rupee-denominated and domestically held, so it cannot suffer the foreign-currency default that felled Sri Lanka (2022), Argentina, or Asia 1997. India owes rupees it can, in extremis, print (M7.04, "denomination is destiny"). The pressure valve is inflation/depreciation, not default, and the real constraint is the interest bill crowding out growth spending (about 40% of the centre's revenue receipts), not solvency. Never map a debt ratio to crisis risk without first asking in whose currency, and held by whom.

Mistake 2: "Unemployment is ~4%, so jobs aren't India's problem." The headline rate is real but misleading in a poor country: distress shows up as informality (~80–90% of the workforce), underemployment, and low participation, especially the very low female LFPR, not as open unemployment, which most Indians cannot afford. The expert reads the quality and formality of jobs and the participation rate, not the unemployment rate, and knows this is the variable that decides whether the demographic bulge becomes a dividend or a burden (§4.4–4.5).

Mistake 3: "The RBI targets core inflation (or the WPI)." No. The RBI's FIT target is 4% headline CPI, food and fuel included, a deliberate and much-debated choice, because Indian inflation expectations are set by the volatile food-and-fuel prices households actually see (food is ~46% of the basket). The WPI is a separate index and not the target (M7.03). Getting the target variable wrong means mis-reading every RBI decision.

Mistake 4: "A managed float means the RBI defends a rupee level." The RBI explicitly has no target level; it intervenes to smooth volatility, not to peg. The expert knows why: spending reserves to defend a level is how you lose the reserves (Bank of England 1992; countless EMs). The buffer's job is deterrence and shock-absorption, and letting the rupee find its (structurally depreciating, PPP-driven) level is the point of choosing a float in the impossible trinity (M7.06).

Mistake 5: "FPI outflows will crash the Indian market like 2013." Largely obsolete, and missing the single biggest structural change: the domestic SIP/DII bid now routinely absorbs foreign selling (§4.4), so 2022–2025 saw heavy FPI outflows with the market barely falling, unthinkable in 2013. The expert holds both the change (real resilience) and the caveat (a sticky retail bid that has never lived through a long bear market is untested). Structural, falsifiable — not a mood.

Mistake 6: "India is the next China." Half-true and dangerous as a model. India's engine is domestic-demand-and-services-led, not the export-and-manufacturing-led model that drove China's mass job-creation, so you cannot assume the same manufacturing-jobs escalator (§4.4), and India's growth is steadier but structurally slower to industrialise. The expert reasons from India's actual structure, not a China analogy, and remembers India trades at a premium valuation the analogy ignores.

Mistake 7: reading the dashboard in levels. The beginner says "CPI is 4.5%, GDP is 6%." The market has already priced those levels; it moves only on rate-of-change (direction and acceleration) and surprise (print versus consensus). "CPI 4.5%" is bullish for bonds if 5.0% was feared and bearish if 4.0% was hoped: same number, opposite reaction (§4.9). The single most important habit-shift here: read every indicator as a delta and a surprise, never a point.

Mistake 8: confusing "great economy" with "great investment." Correct macro is not a buy signal. India is a famously expensive market, and the premium multiple already embeds much of the growth optimism, so a right bull thesis can still be a poor entry (§4.1). The expert lets macro literacy inform the discount rate and the demand outlook, then hands the decision to valuation discipline (Phase 3): the price must leave a margin of safety even if the macro story is true.

Mistake 9: sloppy aggregates. Quoting the central deficit (~4.4%) as if it were the combined (~7–8%); citing headline reserves without netting the RBI's forward book; or comparing an India equity return in rupees to a US return in dollars without adjusting for the ~2–4%/yr currency drift. The expert always states which aggregate, net of what, and in which currency. Sloppiness is where wrong conclusions hide.

Worked examples

Worked Example 1: The oil shock and the twin-deficit-via-oil channel (India)

Illustrative numbers, ~FY25/26 scale; pull the current oil-sensitivity and GDP figures before using in real work.

Setup. Sustained crude oil rises from $70 to $100/barrel, a +$30 shock. India's nominal GDP ≈ $3.9 trillion. Rule of thumb (it moves with import volumes): every +$10/bbl sustained adds ≈ $14bn to the annual import bill and ≈ 0.35% of GDP to the current-account deficit. Fuel is directly ~6–8% of the CPI basket, with meaningful second-round effects. The government is deciding whether to absorb the pump-price rise.

Step 1. The current-account channel. +$30/bbl = 3 × the $10 unit. Extra import bill ≈ 3 × $14bn = ~$42bn. As a share of GDP: $42bn / $3,900bn = ~1.1% of GDP added to the CAD. If the CAD was ~1.2% of GDP, it widens toward ~2.3%, approaching the zone where financing it becomes a live concern.

Step 2. The inflation channel. A ~43% oil price rise, with fuel ~6–8% of CPI and second-round pass-through into transport and input costs, plausibly adds ~0.5–0.9 percentage points to headline CPI over a few quarters (illustrative). If CPI was at 4.0%, this could push it toward/through the upper half of the 2–6% band, constraining the RBI (it can no longer cut, and may have to signal a hold or hike), which raises the discount rate under every Indian valuation.

Step 3. The fiscal channel. If the government cushions retail fuel prices (excise cuts / subsidy) to protect households and inflation, it widens the fiscal deficit by the amount of the cushion, say ~0.2–0.4% of GDP. If instead it passes the price through, it protects the fiscal but hits consumption and growth. Either path has a cost.

Step 4. The synthesis (the triple whammy). One shock lights up three parts of the dashboard at once: CAD wider (~+1.1% GDP), inflation higher (~+0.5–0.9pp), fiscal pressured, and (because the CAD must be financed) downward pressure on the rupee, which is itself inflationary, a feedback loop. This is why oil is India's master external vulnerability, and why "what is oil doing?" is the first question in any India macro scenario. The mirror image is a clean tailwind: a sustained oil fall narrows the CAD, cools inflation, frees fiscal room, and firms the rupee, an unambiguous upgrade to the India macro read.

Worked Example 2: Judging reserve adequacy (India)

Illustrative numbers; pull live figures from the RBI Weekly Statistical Supplement and the external-debt release.

Setup. FX reserves = $690bn. Annual imports = $720bn (so average monthly imports = $60bn). Short-term external debt due within 12 months = $130bn. The RBI's net short forward book = $60bn.

Step 1. Months of import cover. = reserves ÷ monthly imports = $690bn ÷ $60bn = 11.5 months. The IMF rule of thumb is ≥3 months; a comfortable EM sits well above. 11.5 months is very comfortable, a deep buffer.

Step 2. Reserves-to-short-term-external-debt (Guidotti–Greenspan). = $690bn ÷ $130bn = 5.3×, i.e. reserves cover 530% of short-term external debt. The rule wants ≥100% (survive a full year of zero rollover). India covers it more than five times over, the direct lesson of 1991 and 2013, institutionalised.

Step 3. Net of the forward book (the sophisticated adjustment). Headline reserves overstate the deployable buffer by the RBI's net short forward position. Adjusted reserves ≈ $690bn − $60bn = $630bn, giving import cover of $630 ÷ $60 = 10.5 months and short-term-debt cover of $630 ÷ $130 = 4.8×. Still very comfortable, but the honest number is lower than the headline.

Step 4. The judgment. On every metric India is comfortably adequate to abundant: a deterrent buffer that makes a speculative attack not worth trying, not a fund to be spent defending a level. This is a core pillar of the §4.7 "resilient, not fragile" conclusion, and it is why a large debt ratio (§4.6) and a structural CAD (§4.7) do not add up to a crisis: the external buffer is deep and the debt is in rupees.

Worked Example 3: Reading a regime from a dashboard snapshot

Setup. An illustrative month's dashboard (levels, prior, and consensus given so you can compute change and surprise):

Step 1. Read as change + surprise. Growth: PMI decelerating (53.2→50.5) and a negative surprise (below 52.0 consensus); retail sales decelerating and below consensus; claims rising → growth is rolling over. Inflation: CPI falling (4.1→3.4) and a downside surprise (below 3.6) → disinflation, faster than expected. Policy: central bank paused with a dovish tilt → pivoting toward easing. Financial conditions: the curve is re-steepening from inversion (−30→−5, a bull steepener) and HY spreads are widening → the classic late-cycle-into-recession signature. Global: oil soft (helps), rupee stable, but VIX rising → risk sentiment deteriorating.

Step 2. Name the regime (2×2). Growth decelerating + inflation decelerating = the disinflationary slowdown quadrant (bottom-left), heading toward the recession quadrant. Two-word call: "disinflationary slowdown."

Step 3. Place the debt-cycle stage. The re-steepening from a prior inversion, widening credit spreads, softening labour, and a central bank pivoting dovish after a hiking cycle are the hallmarks of late-cycle turning into early-recession/deleveraging. (The inversion already happened; the un-inversion as growth rolls over is historically the more reliable recession trigger than the inversion itself.)

Step 4. Draw the stance (with triggers). Regime + stage ⇒ a risk-off tilt: favour long-duration government bonds (both growth and inflation falling is the bond-bullish quadrant, and the dovish pivot confirms it); tilt equities toward quality/defensives over cyclicals; raise some cash; be cautious on credit as spreads widen. The disinflation-plus-dovish-pivot is eventually bullish for long-duration equities once the growth scare is priced — but not yet, with claims still rising and spreads still widening. Falsification triggers: this read flips if (a) PMI/claims re-accelerate (growth stabilises → back toward Goldilocks), or (b) the next two CPI prints surprise high (→ stagflation risk, which would kill the long-duration-bond call). Named triggers turn the call into a falsifiable thesis you update as the data land.

Worked Example 4: India's debt dynamics: why the scary ratio is stable (r vs g)

Applies the M7.04 debt-dynamics arithmetic to India. Illustrative numbers.

Setup. General-government debt = 82% of GDP. Nominal GDP growth g = 10.5%. Effective interest rate on the debt r = 7.2%. Primary deficit (deficit excluding interest) = 1.5% of GDP. The debt-ratio change is:

Δ(debt/GDP) = [(r − g) / (1 + g)] × d_prev + primary deficit (all as fractions; primary deficit positive)

Step 1. The automatic (r − g) term. (r − g) = 0.072 − 0.105 = −0.033. Divided by (1 + g) = 1.105 → −0.02986. Times d_prev = 0.82 → −0.0245, i.e. the growth-versus-interest dynamic shrinks the debt ratio by 2.45% of GDP this year, purely because the economy grows nominally faster than the debt accrues interest.

Step 2. Add the primary deficit. Δ = −0.0245 + 0.015 = −0.0095 → the debt ratio falls ~0.95 percentage points, from 82.0% to ~81.05%, even while running a 1.5%-of-GDP primary deficit. This is India's fiscal saving grace: fast nominal growth lets it "grow out of" the debt despite persistent deficits.

Step 3. The stress test (why it isn't guaranteed). Suppose growth slows so g falls to 6% (below r = 7.2%). Now (r − g)/(1 + g) = (0.072 − 0.060)/1.06 = +0.01132; × 0.82 = +0.00928; + 0.015 primary deficit = +0.024 → the debt ratio rises 2.4 points instead of falling. The entire resilience rests on nominal g staying above r. So the analyst's fiscal watch-item is not the debt level but the growth-versus-interest-rate gap: as long as India grows nominally faster than it borrows, the scary ratio erodes; if growth stalls while rates stay high, the arithmetic turns against it. This is the precise, falsifiable version of "India can carry high debt": true conditional on r < g.

Worked Example 5: The twin deficits, sized in rupees (India)

Illustrative, ~FY25/26 scale; pull the live Union Budget, the RBI balance-of-payments release, and the nominal-GDP print before using in real work.

Setup. Nominal GDP ≈ ₹340 lakh crore (≈ $3.9 trillion at ₹87/$). On the fiscal side the central government runs a deficit of 4.4% of GDP and the states a combined 2.9%. On the external side, the balance of payments (annualised, $bn) reads: goods trade deficit −$300, services surplus +$170, net remittances (secondary income) +$125, primary income & other net −$42. The "twin deficits" are the fiscal deficit and the current-account deficit together: the two gaps that must both be financed, and that come under pressure at the same moment (§4.7).

Step 1. The fiscal deficit, and the aggregate that matters. Central = 4.4% × ₹340 lc = ₹14.96 lakh crore. States = 2.9% × ₹340 lc = ₹9.86 lakh crore. The general-government (combined) deficit is the sum: 4.4% + 2.9% = 7.3% of GDP (₹24.82 lakh crore). Quoting the central 4.4% as if it were the whole is the classic aggregate trap (Mistake 9). India is a federation, and the figure that measures the state's total draw on the economy's savings is the combined 7.3%.

Step 2. The current-account deficit, from its components. CAD = −300 + 170 + 125 − 42 = −$47bn. As a share of GDP: $47bn ÷ $3,908bn = 1.20% of GDP, a modest, structural deficit. Notice what makes it modest: the goods deficit alone is $300bn (7.7% of GDP); the services surplus + remittances ($295bn) claw almost all of it back. India's invisibles, the IT/BPO export machine and the world's largest remittance inflow, are the reason a huge goods gap nets to a ~1% CAD. Take either lever away and the external picture changes completely.

Step 3. The twin, and who finances which. Side by side: fiscal 7.3% of GDP, CAD 1.20% of GDP. But the two are financed in different currencies, and that is the whole §4.6 point. The fiscal deficit is funded ~95%+ in rupees, domestically (SLR-bound banks, insurers, the RBI), with no foreign-currency-default failure mode. The CAD is a dollar gap that must be covered by the capital account (FDI + FPI + external borrowing), so it is where the real external vulnerability lives. In rupees, the $47bn CAD is ≈ ₹4.09 lakh crore of net foreign financing needed this year.

Step 4. The analyst's read. A generic-EM lens adds "7.3% + 1.2% = 8.5% of twin-deficit danger" and panics. The trained lens separates by denomination: the large fiscal deficit is a rupee problem (its cost is the interest bill and the discipline on RBI independence, not default, §4.6), while the small CAD is the dollar problem (its cushion is the reserve buffer plus stable FDI and bond-index flows, §4.7). Same two numbers, a completely different risk map, which is exactly why Mistake 1 and Mistake 9 are the two most common India-macro errors.

gdp_lc = 340.0                         # nominal GDP, Rs lakh crore
central, states = 0.044*gdp_lc, 0.029*gdp_lc
general = central + states             # 24.82 lakh crore
gdp_usd_bn = gdp_lc*1e5*1e7/87/1e9     # = 3908  ($bn)
cad = -300 + 170 + 125 - 42            # = -47   ($bn)
print(round(general,2), round(general/gdp_lc*100,1))            # 24.82 7.3
print(cad, round(abs(cad)/gdp_usd_bn*100,2))                    # -47 1.2
print(round(300/gdp_usd_bn*100,1), round(abs(cad)*87/1000,2))   # 7.7 4.09

Worked Example 6: Real policy rates, and why India's "high" rate is no free lunch (US comparison)

Illustrative, synthetic rates chosen for the arithmetic, not dated market prints; verify the live repo, Fed-funds and CPI prints before drawing a conclusion about today.

Setup. The RBI repo rate is 5.25% with Indian CPI at 4.0%. The US federal funds rate is 4.25% with US CPI at 2.5%. A dollar-based investor notices India pays a full percentage point more and wonders whether to move cash into Indian instruments. Two questions: which central bank actually runs the tighter policy, and does the nominal pickup survive the currency?

Step 1. The real US policy rate. Real ≈ nominal − inflation = 4.25 − 2.5 = 1.75% (Fisher-exact: 1.0425/1.025 − 1 = 1.71%).

Step 2. The real India policy rate. 5.25 − 4.0 = 1.25% (exact: 1.0525/1.04 − 1 = 1.20%).

Step 3. The counterintuitive result. India's nominal rate is higher (5.25% vs 4.25%), yet its real rate is lower (≈1.20% vs ≈1.71%), because Indian inflation is higher. The US is running the tighter real stance. Never compare nominal policy rates across countries: a higher number can be looser money once you deflate it (M0.03's real-vs-nominal law, and why the distinction bites harder for India than for the US).

Step 4. The cross-border hurdle. By relative PPP, the currency of the higher-inflation country drifts down to keep traded-goods prices aligned: expected rupee depreciation ≈ India CPI − US CPI = 4.0 − 2.5 = ~1.5%/yr (exact 1.04/1.025 − 1 = 1.46%). So a dollar investor earning India's 5.25% nominal expects to lose ~1.5% to the falling rupee, netting 1.0525 × 1.025 ÷ 1.04 − 1 = ~3.7% in dollar terms, below the 4.25% available at home. The "high" Indian rate was never free money; the currency drag is the toll (§4.3). This is why a dollar investor demands a higher hurdle on Indian assets, why India's equity returns look better in rupees than in dollars, and why you must never quote an India return without its currency.

us_real  = 1.0425/1.025 - 1              # 1.71%
in_real  = 1.0525/1.04  - 1              # 1.20%
exp_depr = 1.04/1.025   - 1              # 1.46%  (rupee, relative PPP)
usd_from_india = 1.0525*1.025/1.04 - 1   # 3.73% vs 4.25% at home
print(round(us_real*100,2), round(in_real*100,2))          # 1.71 1.2
print(round(exp_depr*100,2), round(usd_from_india*100,2))  # 1.46 3.73

Practice set

Problems 1–4 are guided (framework applied with scaffolding), 5–9 independent, 10–12 exam-condition (12 minutes each, closed book). All figures are "illustrative, ~mid-2026 scale — verify"; use the numbers given. Write your full answer before reading the solution.

P1 (guided: the corridor and the pre-committed rupee). The RBI repo rate is 5.50% with a symmetric ±25-bp corridor; CRR is 3.0% and SLR is 18% (the cash reserve ratio moves, and it reached 3.0% when the phased 2025 cuts completed; check against the current RBI release before quoting it). (a) State the SDF and MSF rates and which is floor/ceiling. (b) What is the corridor width? (c) Roughly what fraction of a bank's deposits is pre-committed before it can make a commercial loan, and why does this cap the credit-deposit ratio below 100%?

Solution. (a) SDF = 5.25% (floor: the rate a bank earns parking spare reserves) and MSF = 5.75% (ceiling: the rate a bank pays when short); the repo (5.50%) sits between them. (b) Width = MSF − SDF = 5.75 − 5.25 = 50 bps. (c) CRR + SLR = 3.0% + 18% = 21.0% of NDTL is locked up (cash at the RBI plus government securities) before commercial lending, which is exactly why the Indian CD ratio's comfort zone is ~75–85%, not 100% (M5.01), a structural ceiling US banks don't face.

P2 (guided: read the MPC). The MPC votes 4–2 to hold the repo, shifts the stance from "withdrawal of accommodation" to "neutral," and the two dissenters wanted a cut. Consensus expected a hold with no stance change. Classify the overall signal (hawkish/dovish) and explain what an analyst should conclude about the next move.

Solution. Dovish, despite the unchanged rate. Three dovish tells: the stance shift to "neutral" (removing the tightening bias, since a stance change often out-signals the rate decision); the dissents for a cut (a dovish minority that could become a majority: the split is information); and the fact that the stance change was not expected (the surprise is what moves the market). Conclusion: the tightening cycle is over and the next move is more likely a cut than a hike. The 10-year G-sec yield should fall on this, and the discount rate under Indian equities eases at the margin.

P3 (guided: reserves adequacy). Reserves = $660bn; annual imports = $792bn; short-term external debt (≤12m) = $120bn. (a) Compute months of import cover. (b) Compute reserves-to-short-term-debt. (c) Is India externally adequate, and what is the buffer for?

Solution. (a) Monthly imports = 792/12 = $66bn; import cover = 660/66 = 10.0 months (≥3 rule → very comfortable). (b) 660/120 = 5.5× = 550% cover of short-term external debt (≥100% Guidotti–Greenspan rule → abundant). (c) Comfortably adequate. The buffer is a deterrent and shock-absorber: it makes a speculative attack not worth trying and smooths genuine disorder. It is not a fund to be spent defending a rupee level (spending it to defend a level is how you lose it).

P4 (guided: the 2×2). For each, name the growth–inflation quadrant and one characteristic asset winner: (a) PMIs accelerating, CPI falling; (b) growth slowing, CPI accelerating; (c) both growth and inflation falling.

Solution. (a) Goldilocks / recovery (growth up, inflation down) → risk-on equities, especially growth/long-duration; the best quadrant for stocks. (b) Stagflation (growth down, inflation up) → commodities/gold and cash; punishes both nominal bonds and most equities, the hardest quadrant. (c) Disinflationary slowdown → recession (both falling) → long-duration government bonds and quality/defensive equities; the classic risk-off.

P5 (independent: the oil channel). Crude rises a sustained +$20/bbl. Using the rule of thumb (+$10/bbl ≈ +$14bn import bill ≈ +0.35% of GDP on the CAD), (a) estimate the CAD impact, and (b) name all three channels through which the shock hits India.

Solution. (a) +$20 = 2 units → extra import bill ≈ 2 × $14bn = ~$28bn, and CAD impact ≈ 2 × 0.35% = ~0.7% of GDP wider. (b) The triple whammy: (1) current account, the wider import bill; (2) inflation, since fuel feeds CPI directly and via input costs, threatening the 4% target and constraining the RBI; (3) fiscal, because cushioning pump prices widens the fiscal deficit while passing them through hits consumption/growth. Plus the second-round: a wider CAD pressures the rupee, which is itself inflationary, a feedback loop. Oil is India's master external vulnerability.

P6 (independent: denomination is destiny). A commentator says "India's general-government debt is ~83% of GDP, higher than Sri Lanka's before it defaulted — India is next." Rebut this in three sentences, and state the constraint that is real.

Solution. India's public debt is ~95%+ rupee-denominated and held domestically (by SLR-bound banks, insurers, pension funds, the RBI), whereas the crises that cause sovereign default (Sri Lanka 2022, Argentina, Asia 1997) are foreign-currency debt crises where a country cannot print what it owes. Because India owes rupees it can, in extremis, print, its failure mode is inflation/depreciation, not default (M7.04, "denomination is destiny"), so the debt ratio simply does not carry the same crisis risk. The constraint that is real is the interest bill: over a third of the centre's revenue receipts goes to debt service, crowding out the growth-enhancing spending the economy needs, plus the discipline it imposes on RBI independence (fiscal-dominance risk).

P7 (independent: the domestic cushion). In 2013 heavy FPI outflows crashed the Indian market and the rupee; in 2022–2025 heavy FPI outflows barely moved it. (a) What structural change explains this? (b) State the caveat.

Solution. (a) The financialisation of household savings created a large, sticky, domestic bid: monthly SIP inflows into equity mutual funds (~₹26,000–27,000 cr/month, illustrative) mean domestic institutional investors (DIIs) now routinely absorb FPI selling almost rupee-for-rupee, a market-structure change that simply did not exist in 2013. (b) The caveat: this sticky retail bid has only ever known a rising market and is untested in a long, grinding bear market. The open behavioural question is whether SIP flows keep coming if the market falls 30% and stays down for two years. A genuine buffer, with an asterisk.

P8 (independent: change vs surprise). CPI prints 4.5%. In Scenario A the consensus was 5.0%; in Scenario B it was 4.0%. Predict the bond-market reaction in each and explain the principle.

Solution. Scenario A: bond yields fall (bonds rally), because inflation came in below what was feared, a dovish surprise that pulls forward rate-cut expectations. Scenario B: bond yields rise (bonds sell off), because inflation came in above what was hoped, a hawkish surprise. Same 4.5% print, opposite reactions. The principle: markets have already priced the expected level, so they move only on the gap between print and expectation (the surprise), not on the level itself (M7.05). This is why the dashboard must be read versus-expectations, and why you must know both the number and what was expected.

P9 (independent: the resilience buffers). List the specific things that changed between the 2013 taper tantrum (when India was a "Fragile Five" EM) and today that make India externally resilient.

Solution. Five specifics: (1) Inflation targeting adopted (2016), which anchored previously-unmoored inflation and expectations (the biggest institutional upgrade); (2) the CAD narrowed from ~4.8% of GDP to ~1–1.5%, removing the financing gap; (3) FX reserves roughly doubled to ~$650–700bn (~10–11 months of import cover vs ~7 in 2013), a deep deterrent buffer; (4) real interest rates turned positive, so holding rupees pays, reducing flight incentive; (5) the financing mix improved, with more stable FDI and now passive bond-index money, plus the domestic SIP/DII bid under equities. Net: relatively resilient, not fragile, though not immune to a large enough oil-plus-risk-off shock.

P10 (exam, 12 min: read the regime). A month's dashboard: Manufacturing PMI 55.8 (prior 54.0, consensus 54.5); retail sales +6% YoY (prior +4%); wage growth accelerating; CPI 5.4% (prior 4.6%, consensus 5.0%); the central bank just hiked 25 bps with a hawkish tone; 10y–2y curve flattening toward zero; credit spreads tight; oil rising; VIX low. (a) Read as change + surprise. (b) Name the quadrant. (c) Place the debt-cycle stage. (d) Give the stance and two falsification triggers.

Solution. (a) Growth: PMI accelerating (54.0→55.8) and a positive surprise; retail and wages accelerating → growth strong and firming. Inflation: CPI accelerating (4.6→5.4) and an upside surprise (above 5.0) → inflation rising, hotter than expected. Policy: hiking, hawkish → tightening. Conditions: curve flattening, spreads complacently tight, oil rising, VIX low. (b) Growth up + inflation up = the reflation / overheat quadrant (top-right). (c) Late cycle / boom: accelerating growth and inflation, a central bank hiking into it, a flattening curve, and complacent spreads are the classic late-cycle top signature. (d) Stance: favour real assets, commodities, value/cyclicals, and inflation-linked bonds over long-duration growth equities and nominal bonds (which the hiking/inflation combo punishes); begin to de-risk as the flattening curve warns the top is near; hold some cash for the turn. Triggers: the read flips (i) toward stagflation if growth surprises down while inflation stays hot (cut cyclicals), or (ii) toward slowdown, de-risking hard, if the curve inverts and PMIs roll over (the late-cycle top confirmed).

P11 (exam, 12 min: the sources). For each figure, name the primary source and (where relevant) the release timing: (a) India's CPI; (b) the IIP; (c) quarterly real GDP; (d) the central fiscal deficit and borrowing plan; (e) FX reserves; (f) the official unemployment rate; (g) a plain-English decode of the Union Budget.

Solution. (a) CPI: MOSPI/NSO, released ~12th of each month for the prior month (the RBI's target variable). (b) IIP: MOSPI/NSO, ~last working day of the month, for the month ~6 weeks prior (it lags). (c) Real GDP: MOSPI/NSO national accounts, quarterly (end-Feb/May/Aug/Nov). (d) Fiscal deficit & borrowing: the Union Budget (Ministry of Finance), delivered Feb 1 (with the Economic Survey the day before). (e) FX reserves: RBI, in the Weekly Statistical Supplement (WSS) each Friday (via DBIE / the RBI Data Warehouse). (f) Unemployment: MOSPI/NSO's PLFS (official; annual with quarterly urban rounds); CMIE's CPHS gives a higher-frequency monthly read (paid, but widely reported). (g) PRS Legislative Research (prsindia.org), its annual "Union Budget Analysis" (free, non-partisan).

P12 (exam, 12 min: the full synthesis). Trace, step by step, what happens to India when a sustained oil spike coincides with a Fed hiking cycle (a global risk-off). Cover: the CAD, the rupee, inflation, the RBI's response, the fiscal, FPI flows, and the implication for Indian equities and bonds. Then state the one structural fact that stops this from becoming a 2013-style crisis.

Solution. The chain: (1) CAD widens: the oil import bill jumps (§4.7 rule of thumb), and a Fed-hiking risk-off simultaneously pulls FPI money out of Indian bonds/equities, so the deficit is both bigger and harder to finance. (2) The rupee falls: a wider CAD plus capital outflow plus a stronger dollar (DXY up on Fed hikes) all push USD/INR up; the RBI sells dollars from reserves to smooth the fall, which drains rupee liquidity (a tightening) unless sterilised. (3) Inflation rises: directly from oil and via the weaker rupee (imported inflation), threatening the 4% target. (4) The RBI is boxed in: it may have to hold or hike to defend the rupee and inflation even as growth slows, a painful trade-off. (5) The fiscal: if the government cushions fuel prices, the deficit widens; the twin deficit (fiscal + CAD) is under pressure at once. (6) Markets: the rising discount rate and risk-off hit long-duration equities hardest; bond yields rise (falling prices) as inflation and the fiscal worsen and foreign holders sell. The one structural fact that prevents a 2013-repeat crisis: the deep FX-reserve buffer (~10–11 months of import cover, §4.3) plus the domestic SIP/DII bid (§4.4). Reserves absorb the currency pressure without India running out, and domestic flows cushion the equity outflow, so the shock is painful but not catastrophic, and India's rupee/domestic debt means it cannot be pushed into a sovereign-default crisis (§4.6). Resilient, not immune.

P13 (guided: the current account from its components). India's balance of payments (annualised, $bn): goods trade deficit −$310, services surplus +$165, net remittances +$120, primary income & other net −$25. Nominal GDP ≈ $3.9 trillion. (a) Compute the current-account balance ($bn) and as a % of GDP. (b) What must finance it? (c) The goods deficit alone is ~7.9% of GDP. What turns that into a ~1% CAD?

Solution. (a) CAD = −310 + 165 + 120 − 25 = −$50bn; as a share of GDP, 50 ÷ 3,900 = 1.28% of GDP, a modest, structural deficit. (b) A CAD is a dollar gap; it must be financed by the capital account: FDI (stable), FPI (flighty), and external commercial borrowing. (c) The invisibles: the services surplus + remittances ($285bn) offset almost the entire $310bn goods deficit. India's IT/BPO export machine and its world-largest remittance inflow are precisely why a large goods gap nets to a small CAD, the structural reason it is modest (§4.7).

P14 (guided: the real policy rate, across two countries). The RBI repo is 5.5% with Indian CPI at 4.5%; the US federal funds rate is 4.0% with US CPI at 2.8%. (a) Compute each real policy rate (nominal − inflation). (b) Which country runs the tighter real stance? (c) What is the one-line lesson?

Solution. (a) India real ≈ 5.5 − 4.5 = 1.0% (exact 1.055/1.045 − 1 = 0.96%); US real ≈ 4.0 − 2.8 = 1.2% (exact 1.04/1.028 − 1 = 1.17%). (b) The US: its real rate (~1.2%) is higher than India's (~1.0%) despite a lower nominal rate, because US inflation is lower. (c) Never compare nominal rates across countries; deflate first. A higher nominal rate can be the looser real stance, the same trap that makes a dollar investor's "pickup" from Indian rates an illusion once the currency is netted (Worked Example 6).

P15 (independent: the fiscal aggregates and the real constraint). Nominal GDP ≈ ₹340 lakh crore. The central deficit is ₹15.3 lakh crore; the states' combined deficit is ₹9.5 lakh crore. Interest on central debt is ₹11.6 lakh crore against central revenue receipts of ₹31 lakh crore. (a) Express the central, state, and general-government deficits as % of GDP. (b) Compute the interest-to-revenue ratio. (c) Which number is the binding fiscal constraint, and why is it not default risk?

Solution. (a) Central 15.3/340 = 4.5%; states 9.5/340 = 2.8%; general (combined) 24.8/340 = 7.3% of GDP. Quote the 7.3%, not the 4.5%, when you mean the economy's total fiscal draw (Mistake 9). (b) Interest-to-revenue = 11.6/31 = 37.4%: over a third of every revenue rupee is pre-committed to servicing debt before a rupee reaches roads, schools, or skilling. (c) The binding constraint is the interest bill (~37% of revenue) crowding out growth spending, not solvency. India's debt is ~95%+ rupee-denominated and domestically held, so it cannot suffer a foreign-currency default (denomination is destiny, §4.6); the failure mode is inflation/depreciation and the opportunity cost of the interest bill, not a Sri-Lanka-style default.

P16 (independent: reserve adequacy, net of the forward book). FX reserves = $650bn; annual imports = $780bn; short-term external debt (≤12m) = $135bn; the RBI's net short forward book = $50bn. (a) Months of import cover and reserves-to-short-term-debt on the headline number. (b) Both metrics net of the forward book. (c) Verdict, and what the buffer is for.

Solution. (a) Monthly imports = 780/12 = $65bn → import cover = 650/65 = 10.0 months; reserves/ST-debt = 650/135 = 4.8× (481%). (b) Deployable reserves = 650 − 50 = $600bn → cover = 600/65 = 9.2 months; 600/135 = 4.4×. The headline overstates the true buffer by the net forward position, the sophisticated adjustment of §4.3. (c) Comfortably adequate to abundant on every metric (import cover ≫3, ST-debt cover ≫100%). The buffer is a deterrent and shock-absorber: it makes a speculative attack not worth trying and smooths genuine disorder. It is not a fund to be spent defending a rupee level (spending it to defend a level is how you lose it).

P17 (independent: the macro→valuation transmission). A dovish MPC cut pulls the 10-year G-sec (the risk-free rate) down from 7.0% to 6.6% (−40 bps). An Indian equity has a cost of equity Ke = Rf + β × ERP, with β = 1.2 and an equity risk premium of 6.0%. (a) Compute Ke before and after. (b) With cash flows unchanged, the % change in fair value if the stock is a no-growth perpetuity (V = CF/Ke). (c) The same, if it is a Gordon grower at g = 6% (V = CF₁/(Ke − g)). (d) What does the gap between (b) and (c) tell you about which stocks are most rate-sensitive?

Solution. (a) Before: Ke = 7.0 + 1.2 × 6.0 = 14.2%; after: 6.6 + 7.2 = 13.8%. (b) No-growth: value ∝ 1/Ke, so % change = 14.2/13.8 − 1 = +2.9%. (c) Gordon: the denominator falls from 14.2 − 6 = 8.2% to 13.8 − 6 = 7.8%, so % change = 8.2/7.8 − 1 = +5.1%. (d) A 40-bp fall in the risk-free rate, with cash flows untouched, lifts fair value 2.9–5.1% purely through the discount rate, and the higher-growth (longer-duration) stock re-rates more (5.1% > 2.9%), because more of its value sits in distant cash flows discounted at the now-lower rate. This is the exact channel M8's macro-overlay line captures: the repo → G-sec → discount-rate path is how the macro weather reaches a single business's price, the loop the whole program closes on.

P18 (exam, 12 min: stretch synthesis, all four parts). Illustrative, ~mid-2026 scale. (i) A sustained +$25/bbl oil spike; rule of thumb +$10/bbl ≈ +$14bn import bill ≈ +0.35% of GDP on the CAD: estimate both impacts. (ii) Repo 5.25%, CPI 4.2%: the real repo rate. (iii) General-government debt d = 80%, nominal growth g = 9%, effective rate r = 7.5%, primary deficit 2.0% of GDP; using Δ(debt/GDP) = [(r−g)/(1+g)]×d + primary deficit, compute the change. (iv) After the oil pass-through, PMIs are rolling over while CPI surprises higher: name the growth–inflation quadrant.

Solution. (i) 2.5 units → import bill ≈ 2.5 × $14bn = +$35bn; CAD ≈ 2.5 × 0.35 = +0.88% of GDP wider. (ii) Real repo ≈ 5.25 − 4.2 = 1.05% (exact 1.0525/1.042 − 1 = 1.01%). (iii) (r−g)/(1+g) = (0.075 − 0.09)/1.09 = −0.01376; × 0.80 = −0.01101; + 0.020 = +0.009 → the debt ratio rises ~0.9 pp. The nuance: g > r is favourable (the automatic term is negative), but a 2.0% primary deficit outweighs it, so the ratio still climbs. g > r is necessary, not sufficient, when the primary deficit is large. (iv) Growth decelerating + inflation accelerating (and surprising high) = stagflation, the hardest quadrant, favouring commodities/gold/cash and punishing both nominal bonds and most equities. Scoring: 4/4 in time = fluent; ≤2 = redo Worked Examples 1, 4, and 6 and P13–P17 before the mastery check.

P19 (independent: a revision reverses the read). Three consecutive months of US payrolls print at +185k, +160k and +150k. You write in your memo that hiring is comfortably above the break-even pace of +110k a month that holds the unemployment rate flat, and you place the labour block on the strong side of the board. Two months later the same three months have been revised to +120k, +95k and +88k. (a) Compute the three-month average on the first prints and on the revised data. (b) State what happens to your conclusion. (c) Give the total revision in jobs and as a percentage of the original three-month total. (d) Name the one entry in the vintage column that would have told a reader, at the time, how much weight the original read could bear.

Solution. (a) First prints: (185 + 160 + 150) ÷ 3 = 165.0k a month. Revised: (120 + 95 + 88) ÷ 3 = 101.0k a month. (b) The conclusion reverses. On the first prints hiring ran 55k a month above the 110k break-even and the labour block sat on the strong side; on the revised data it runs 9k a month below break-even, which is a labour market that is no longer keeping up with the growth of the workforce. Nothing about the economy changed between the two readings, only the estimate of it. (c) Total revision = (120 + 95 + 88) − (185 + 160 + 150) = 303 − 495 = −192k jobs, which is −192 ÷ 495 = −38.79% of the original three-month total. (d) The estimate stage: all three prints were first prints, and each was still due two monthly revisions plus the annual benchmark. Recording that in the vintage column would have marked the read as provisional at the moment it was written, which is the whole point of the column. The discipline that follows: on a first-print series, state the conclusion and the revision risk in the same sentence, and prefer the surprise lens, which is measured against a consensus formed on the same vintage, over the change lens, which a revision to the prior month can move on its own.

Applied mini-project: THE CAPSTONE: build the dashboard, write the regime memo

This is the Milestone-M8 capstone and the certification of competency C8. It has two deliverables you assemble yourself from primary sources: (A) the ~18-indicator macro dashboard, and (B) a one-page regime memo that reads it. Budget ~3 hours. Everything you need to source is free and primary. Keep both: the dashboard becomes a standing monthly instrument; the memo is the template for every macro-overlay line in Phase 8 onward.

Deliverable A: build the dashboard from primary sources

Assemble the ~18 indicators into one page (a spreadsheet tab or a single sheet of paper), grouped as in §4.9, using the reusable reference artifacts/macro-dashboard.md for each indicator's exact source, code, and healthy/warning zones. For every indicator record four things: the current level, the prior period, the consensus/expected (where one exists), and your one-word read (accelerating/decelerating/stable, surprise up/down/none). Sourcing:

• Growth: real GDP (US: FRED GDPC1; India: MOSPI national accounts); PMIs (S&P Global / ISM manufacturing & services); IIP (India: MOSPI); retail sales (US: Census / FRED RSAFS; India: proxies such as auto sales and GST).
• Labour: unemployment/payrolls (US: BLS / FRED UNRATE, PAYEMS; India: PLFS + CMIE); jobless claims (US: FRED ICSA); wages (US: FRED CES0500000003).
• Inflation & policy: CPI (US: FRED CPIAUCSL; India: MOSPI, released ~12th); core PCE (US: FRED PCEPILFE); the policy rate & implied path (FRED FEDFUNDS + CME FedWatch; RBI repo + the OIS curve); inflation expectations (FRED T5YIE, MICH; RBI household survey).
• Financial conditions: the yield curve, both 10y–2y (FRED T10Y2Y) and 10y–3m (FRED T10Y3M), plus the India G-sec curve; the 10-year yield (FRED DGS10; India 10y G-sec); credit spreads (FRED BAMLH0A0HYM2 for HY, BAMLC0A0CM for IG); a financial-conditions / lending-standards read (Chicago Fed NFCI; the Fed's SLOOS; RBI's credit conditions).
• Global / FX: DXY and USD/INR (FRED DTWEXBGS; RBI reference rate); oil and gold (FRED DCOILWTICO, GOLDAMGBD228NLBM); equities and the VIX (FRED VIXCLS; Nifty/Sensex).
• India add-ons: the current account/CAD (RBI BoP release, quarterly); FPI flows (NSDL/CDSL, monthly/daily); FX reserves (RBI WSS, Fridays); GST collections (PIB/GST portal, monthly on the 1st); credit growth (RBI, fortnightly); the fiscal deficit vs target (CGA monthly accounts; the Union Budget).

Dashboard-build rubric (self-score; pass ≥9/14 with the vintage row scored, or ≥8/12 on the six original rows):

Deliverable B: write the one-page regime memo

Follow the fixed four-step structure from §4.10. Keep it to one page. The structure, then a full worked model answer.

The memo structure.

1. Indicator read (change + surprise): one short paragraph per group (growth, labour, inflation/policy, financial conditions, global/FX, India), stating the rate-of-change and the surprise, not the levels. Flag conflicts and say which group you weight.
2. Regime call (the 2×2): name the growth–inflation quadrant in two words, with the one-line justification.
3. Debt-cycle stage: early / mid / late / recession-deleveraging, citing the cycle-specific tells (curve, spreads, credit growth, central-bank posture).
4. Portfolio-stance implication + triggers: the tilt (risk-on/off; duration; cyclicals/credit; cash; India-specific rupee/oil/rate sensitivity), and the specific indicators whose change would flip the call.

MODEL ANSWER: a filled regime memo (illustrative; numbers frozen for teaching; do NOT treat as a live read).

REGIME MEMO · [Illustrative month], global + India · one page 1. Indicator read (change + surprise). Growth: Global PMIs holding in modest expansion (52–53) and stable vs prior; US retail resilient, slightly above consensus. India IIP and GST collections accelerating (GST at a record run-rate), auto sales firm → growth solid, India firming, no acceleration or rollover globally. Labour: US payrolls cooling gently (moderating from hot), claims low and stable, wage growth decelerating toward ~4% → labour normalising, not cracking. Inflation & policy: US core PCE falling toward target and slightly below consensus; India CPI benign at ~3.7%, within band and below the RBI's own path. Both central banks easing, with the Fed cutting toward neutral and the RBI having cut to ~5.25% with a neutral stance → disinflation continuing, policy turning supportive. Financial conditions: Curves re-steepening from prior inversion (bull-ish, growth not collapsing); credit spreads tight and stable; financial conditions easing → supportive, mild complacency worth noting. Global/FX: DXY soft, USD/INR stable (~87); oil soft (a clean India tailwind); gold firm; equities near highs, VIX low → benign risk backdrop; low VIX is the one caution flag. India: CAD narrow (~1% of GDP, helped by soft oil); FPI flows turning positive on bond-index inclusion; reserves ample (~$690bn, ~11 months cover); credit growth healthy (not overheated); fiscal on its glide-path. → India externally resilient, domestically firming. Weighting: I weight the leading indicators (PMIs, curves, spreads, surprises) over the lagging (GDP, unemployment); the surprise cluster is mildly dovish/pro-growth. 2. Regime call (2×2): "disinflationary expansion" (Goldilocks-leaning). Growth is steady-to-firming while inflation is falling: the top-left/Goldilocks quadrant. Not a boom (growth isn't accelerating hard) and not a slowdown (no rollover), but a benign, disinflationary, policy-supported expansion. 3. Debt-cycle stage: mid-cycle. Central banks easing from a peak (not yet stimulative), curves re-steepening from inversion, spreads tight, credit growth healthy but not hot: the signature of a mid-cycle setting, past the late-cycle tightening scare, not yet at an early-recovery trough. India specifically sits early-mid: a fresh easing cycle, healthy credit, ample buffers. 4. Stance + triggers. Constructive / risk-on with guardrails. Tilt: favour equities over cash, with room for long-duration/growth as the discount rate eases (the disinflation-plus-cuts combo is the tailwind that helped them); keep credit exposure but respect tight spreads (limited cushion); India constructive: soft oil + easing RBI + resilient externals + the domestic bid is a supportive mix, tempered by India's premium valuation (the macro is good but much is priced, §4.1; demand a margin of safety on any single name). One caution: low VIX + tight spreads = complacency, so this is risk-on with a hand near the exit, not full-throttle. Falsification triggers. This benign call flips if: (a) oil breaks sharply higher (the India triple-whammy lights up across the CAD, inflation and the fiscal, and the RBI's easing bias dies); (b) inflation surprises back up for two prints (→ reflation/overheat; cut duration); (c) the curve re-inverts and PMIs roll over (→ late-cycle-into-slowdown; de-risk, add duration); or (d) credit spreads gap wider (the first sign the mid-cycle calm is ending). I will re-read the board when any trigger fires; absent a trigger, hold the constructive tilt.

Notice what the model memo does: it never forecasts a number; it reads changes and surprises, forces a two-word regime, places the cycle, converts to a tilt (not a trade list), and, the mark of a professional, ends with the specific triggers that would prove it wrong. That is a falsifiable, updatable thesis, not a mood.

Regime-memo grading rubric (self-score; pass ≥11/14):

Passing the capstone requires both deliverables at their bar and the mastery-check exam below. The M8 milestone is "dashboard built from primary sources + regime memo mapping indicators → cycle stage → stance." File both in your knowledge system; the dashboard gets updated monthly forever, and each month's regime memo goes in the decision journal as a dated, falsifiable call you will grade yourself against later.

Case Lab: run case XL-7.3 (The Economic Survey, Read Cold) in the app's Cases tab for this module. You open the latest official Economic Survey (free, indiabudget.gov.in) and extract, cold, the numbers this module runs on: the growth read, the CPI/inflation framing, the fiscal and current-account trajectories, and the three constraints (jobs, manufacturing, human capital). Then map them onto the growth–inflation 2×2 and draft the regime call, self-scored against our key. It turns the primary-source discipline of §4.8 into a timed exercise, and gives you a real dashboard input to file alongside the capstone.

Reading & resources

• Ministry of Finance, the Economic Survey of India (indiabudget.gov.in): the government's flagship annual economic analysis, released the day before the Union Budget. Read the overview chapter each year; it is the single best official narrative of where the Indian economy is and where the government thinks it's going. [Free] [Intermediate]
• RBI: the Monetary Policy Report, MPC resolutions, and the Handbook of Statistics on the Indian Economy (rbi.org.in; DBIE / the RBI Data Warehouse). Your permanent primary source for the repo/corridor, CRR/SLR, reserves, money, credit, and the balance of payments. Read two consecutive MPC resolutions and diff the language. [Free] [Intermediate]
• PRS Legislative Research: the annual "Union Budget Analysis" and state-finance briefs (prsindia.org). The best free, non-partisan, plain-English decode of the Budget and the fiscal numbers. Read it alongside the primary Budget each February. [Free] [Beginner→Intermediate]
• Ray Dalio, Principles for Navigating Big Debt Crises: the debt-cycle framework the dashboard maps onto; his free How the Economic Machine Works video is the 30-minute intuition. The lens behind §4.9's cycle-stage mapping. [Free (video)/Paid (book)] [Intermediate]
• Vijay Joshi, India's Long Road: The Search for Prosperity: the definitive, evenhanded book on India's growth model, its constraints (jobs, manufacturing, human capital), and the reforms needed; the honest counterweight to boosterism. [Paid] [Intermediate→Advanced]
• Raghuram Rajan, I Do What I Do (on the RBI, FIT, and central-bank independence) and Kelkar & Shah, In Service of the Republic (the logic of India's economic reforms and state capacity). Two of the clearest insider accounts of the institutions in §4.2 and §4.6. [Paid] [Intermediate]
• T.N. Ninan, The Turn of the Tortoise: a lucid journalist's synthesis of India's growth story, strengths, and self-imposed limits; the best single-sitting overview of the §4.1 frame. [Paid] [Beginner→Intermediate]
• MOSPI / NSO (mospi.gov.in): the primary release pages for CPI, IIP, GDP, and PLFS; bookmark the release calendar so you read each print the day it lands. [Free] [Intermediate]
• CMIE (cmie.com): the higher-frequency (monthly) unemployment read and the capex/investment-project data; paid, but its headline unemployment number is widely reported and worth knowing. [Paid] [Intermediate]
• FRED (fred.stlouisfed.org): build the global side of your dashboard here; the codes in the mini-project let you plot every US/global series yourself. The habit of plotting beats any summary. [Free] [Beginner]
• Aswath Damodaran, the country-risk-premium data and posts (his NYU page, updated each January): where India's equity risk premium and country risk premium come from for the discount rate (M3.02–M3.03); the bridge from macro to valuation. [Free] [Intermediate]
• Podcasts / ongoing: Odd Lots (Bloomberg) for practitioners arguing live macro in plain language; India-focused newsletters and the RBI/MOSPI release calendars for staying current. Follow the releases, not the noise. [Free] [Intermediate]
• Going deeper: the IMF's latest Article IV Consultation for India (imf.org → Countries → India: the Staff Report and the Selected Issues papers) [Free] [Advanced]. The outside auditor's view of exactly these numbers: the reserve-adequacy (ARA) assessment behind §4.3, the debt-sustainability analysis behind §4.6's r-vs-g, and the external-balance read behind §4.7, each with its methodology shown. Read it against the Economic Survey to see the same economy scored by the government and by the Fund, and note where they disagree.
• Going deeper: Michael Pettis, The Great Rebalancing: Trade, Conflict, and the Perilous Road Ahead for the World Economy (Princeton University Press) [Paid] [Advanced]. The deepest treatment of the savings-investment (S − I) identity under §4.7's current-account deficit and of why an export-led model (China) and a domestic-demand model (India, §4.4) sit at opposite ends of the same global-imbalances arithmetic. Read it to make the CAD and the twin-deficit link mechanical rather than memorised.

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": "The India macro thesis and its constraints, the FIT framework and the managed float, the reserve buffer, the fiscal and external picture, the exact primary sources and their release calendars, the ~18-indicator dashboard read as change and surprise, the growth-inflation 2x2 and the debt-cycle overlay, and the four-step regime memo",
  "use": "This is the module where a tool earns its keep on the build and must be kept out of the read. Three uses on the build side. (1) FETCH-LAYER DRAFTING — 'write a function that pulls this list of FRED series ids into one tidy frame with a date index, retries on failure, and records the fetch timestamp and the series' last-updated date for each'. The last-updated field is not optional; it is what makes the staleness arithmetic below computable rather than assumed. (2) PARSER DRAFTING FOR THE INDIAN HALF, which is where the labour actually is — the MOSPI CPI and IIP releases, the CGA monthly accounts, the PIB GST press release and the NSDL FPI table are PDFs and web tables on fixed layouts, so a tool that drafts the extraction and you verify against the document once is a genuine multiplier. Re-verify whenever the layout changes, which it will. (3) MEMO CHALLENGE, not memo writing. Hand it your completed regime memo and ask it to argue the OPPOSITE quadrant from the same dashboard, and to name which indicator it is weighting differently to get there. If it cannot construct a coherent opposite, your read is probably robust; if it can, you have found the weak joint before the market does. Where it must not go: the read itself. §4.9's method is change-and-surprise on numbers with dates, and a model asked 'what regime are we in' will produce a fluent, confident, undated quadrant call. That output is indistinguishable from a good one and is built on nothing you can audit. Nor may it supply the repo rate, the CPI print, the reserve level, the CAD, or the fiscal deficit — every one of those is free, primary, and dated, and §4.8 exists to tell you exactly where.",
  "tools": ["Chat assistants — Claude, ChatGPT for fetch-layer and parser drafting, and for adversarially challenging a completed regime memo", "Document/PDF AI — for the Economic Survey overview, the Union Budget and CGA monthly accounts, the RBI Monetary Policy Report and Financial Stability Report, and the MOSPI release tables", "Coding assistants — for the dashboard specification, the transformation layer, the staleness ledger and the threshold arithmetic below"],
  "prompt": "Here is my completed regime memo and the dashboard snapshot it was written from. Argue the OPPOSITE quadrant of the growth-inflation 2x2 as persuasively as you can, using only the indicators in this snapshot, and state explicitly which indicators you are weighting more heavily than I did and why. Then list the two cells whose next release would most cleanly settle the disagreement. Do not supply any indicator value that is not in the snapshot I pasted, and do not tell me what the current levels of any of these series are.",
  "verify": "Four checks, and the fourth is the one this capstone is really about. (1) Every cell carries its source, its series id or document page, its release date and its transformation — a dashboard whose provenance is not written down cannot be re-derived next month and is therefore not a dashboard. (2) The transformation is fixed BEFORE you look at the number, because §4.9's change-and-surprise method is only honest if the method predates the print; a series read as year-on-year in a calm month and as three-month-annualised in a turning month is not an indicator. (3) Every threshold is stated in units of the series' own variability rather than in absolute points, and the family-wise false-fire rate of the whole board is computed — the lens below shows that a naive per-indicator threshold fires somewhere on an eighteen-cell board in most months for no reason at all. (4) The memo is graded on process, exactly as §4.10 says: whether the read was defensible from the data as they stood and whether the triggers were specific enough to fire. A memo that turned out right for a reason you did not name is not a good memo, and a memo that turned out wrong for a reason you DID name is.",
  "diy": "The gate is unaided: you compute months of import cover and the Guidotti-Greenspan ratio and adjust both for the forward book, run the debt-dynamics arithmetic and say what it depends on, size the twin deficits and separate them by denomination, convert nominal policy rates to real ones and to a dollar-based return, read a dashboard snapshot as change and surprise, name the quadrant and the cycle stage, and write the four-step memo ending in specific falsification triggers. Nothing in this callout is admissible in that room."
}

Modern Data Analysis

By hand first. You sized the oil shock through all three channels and saw one price light up the whole board; you judged reserve adequacy on two metrics and then netted the forward book, which is the adjustment that separates a trained reader from a headline reader; you ran India's debt dynamics and named the condition the whole result rests on; you separated the twin deficits by denomination; you converted two nominal policy rates into real ones and then into a dollar return and found the "high" Indian rate was lower on both. And you read a dashboard snapshot as change and surprise and wrote the memo. Every one of those stays by hand, because the capstone's graded skill is the read, and no pipeline supplies a read.

Today's workflow. The dashboard stops being a page you fill in and becomes a specification you execute. Four properties make it reproducible, and they are the difference between an instrument and a scrapbook. A fixed series list, addressed by id. Not "the yield curve" but T10Y2Y; not "India CPI" but the MOSPI CPI (Combined) release, with the item table you take it from. An id survives a website redesign; a description does not. A transformation declared in advance, per series. Three-month-on-three-month annualised for the inflation and activity series, level plus three-month change for the financial-conditions series, four-quarter trailing for the external series, all written down once, applied every month, never chosen after seeing the print. A stated cadence, per series, with its publication lag. This is the property everyone skips and it is the one that decides how much your board can actually claim, as the lens below computes. And a threshold per indicator, stated in units of that series' own variability, together with a rule for how many cells must breach before the memo's regime call is reopened, because an eighteen-cell board with naive per-cell triggers is a machine for generating false alarms. Around all four sits one more discipline: record the vintage. Indian GDP and IIP are revised, US GDP is revised three times before it settles, and a dashboard that overwrites last month's value with this month's revision has quietly erased the surprise it was built to measure.

Tools & sources (IN + US). pandas for the panel and the transformation layer, requests for the fetch, numpy and scipy.stats for the thresholds, matplotlib for the one-page board; US: one API call per series. FRED exposes every series by id as JSON with its own last-updated timestamp and, through its ALFRED vintages, the value as it was first published, which is the only way to reconstruct a genuine surprise after the fact. The board's US ids are already named in §4.9 and the mini-project: GDPC1, RSAFS, UNRATE, PAYEMS, ICSA, CES0500000003, CPIAUCSL, PCEPILFE, T5YIE, MICH, FEDFUNDS, T10Y2Y, T10Y3M, DGS10, BAMLH0A0HYM2, BAMLC0A0CM, NFCI, DTWEXBGS, DCOILWTICO, VIXCLS; add SOFR, WALCL, GFDEGDQ188S and USREC, plus the Treasury's daily par yield curve and its Fiscal Data API (which publishes the average interest rate on the debt monthly, the i of the debt-dynamics arithmetic), the New York Fed's ACM term premium and its 10y–3m recession-probability model, the BIS credit-to-GDP gaps and debt-service ratios, and the CBO outlooks. India: a download-and-parse job, and there is no way around it. RBI DBIE (migrating to the RBI Data Warehouse; check the current URL) offers per-table Excel and CSV exports rather than a queryable API; the Weekly Statistical Supplement arrives Friday as a fixed-layout document carrying reserves, money and bank credit; MOSPI publishes CPI around the 12th and IIP with roughly a six-week lag as press releases plus tables; the CGA monthly accounts give the fiscal deficit against budget estimates; PIB carries GST collections on the 1st; NSDL and CDSL publish FPI flows daily; FBIL publishes the USD/INR reference rate and forward premia daily; CCIL carries the G-sec and money-market data; data.gov.in exposes some MOSPI and ministry datasets through an API key, unevenly. So the honest architecture is asymmetric: the US half of the board is a scheduled API job that runs unattended; the India half is a set of parsers against fixed layouts that break when a layout changes and need a human once a month. Budget for that, and let it dictate the India board's size: a smaller set of Indian cells you genuinely maintain beats a longer list you stop updating in March. Paid: CMIE Economic Outlook for the monthly unemployment rate (India has no official monthly equivalent; PLFS is quarterly urban and annual) and for the capex pipeline, CEIC for harmonised long panels, Bloomberg or Refinitiv for consensus expectations. That last one matters more than it looks: §4.9's surprise lens requires a consensus, and for most Indian releases a free, systematic consensus series does not exist. Press previews quote a poll of economists but no free archive of it exists to backtest against. So the India board's surprise column is, in practice, partly your own prior written down before the print, which is a legitimate substitute only if you actually write it down beforehand.

# The dashboard as a specification, and what its thresholds cost — recomputed in-session (R3)
from math import erf, sqrt, comb
BOARD = [   # (name, id / source, frequency days, publication lag days, transform)
    ("10y-2y curve",    "T10Y2Y",                    1,  0, "level, bp"),
    ("10y yield",       "DGS10",                     1,  0, "level + 3m change"),
    ("HY spread",       "BAMLH0A0HYM2",              1,  0, "level + 3m change"),
    ("Broad dollar",    "DTWEXBGS",                  1,  0, "3m change"),
    ("Core PCE",        "PCEPILFE",                 30, 30, "3m/3m annualised"),
    ("India CPI",       "MOSPI release, ~12th",     30, 12, "3m/3m annualised"),
    ("India IIP",       "MOSPI release, ~6wk lag",  30, 42, "3m/3m annualised"),
    ("FX reserves",     "RBI WSS, Friday",           7,  7, "level, net of forward book"),
    ("Current account", "RBI BoP, quarterly",       91, 90, "% of GDP, 4q trailing"),
]
age = lambda F, L: L + F / 2                     # expected age at a random moment
print(f"{min(age(f,l) for _,_,f,l,_ in BOARD):.1f} .. {max(age(f,l) for _,_,f,l,_ in BOARD):.1f} days")
#      0.5 .. 135.5 days
Phi = lambda z: 0.5 * (1 + erf(z / sqrt(2)))
p   = lambda k: 2 * (1 - Phi(k))                 # per-indicator monthly false-fire rate
N   = 18
print(f"{p(2.0):.2%} {1 - (1 - p(2.0))**N:.1%} "
      f"{sum(comb(N,j) * p(2.)**j * (1-p(2.))**(N-j) for j in range(3, N+1)):.2%}")
#      4.55%   56.8%   4.61%    <- one cell at 2 sigma vs a 3-cell cluster rule

Verify. Prove every calculation against the module's own worked figures before the board goes live. Worked Example 1: 3 × $14bn = $42bn, 42/3,900 = 1.077% of GDP, taking a 1.2% CAD to 2.28%, on an oil move of +42.9%. Worked Example 2: import cover 690/60 = 11.5 months, Guidotti-Greenspan 690/130 = 5.31× (531%), and net of the $60bn forward book $630bn giving 10.5 months and 4.85×. Worked Example 4: (0.072 − 0.105)/1.105 × 0.82 = −0.02449, plus a 1.5% primary deficit gives −0.95 pp and a ratio of 81.05%; in the stress case (0.072 − 0.060)/1.06 × 0.82 = +0.00928, plus 1.5% gives +2.43 pp. Worked Example 5: central ₹14.96 and states ₹9.86 lakh crore, general government ₹24.82 lakh crore at 7.3% of GDP; GDP $3,908bn; CAD −$47bn at 1.20% of GDP against a goods gap of 7.68% and invisibles of 6.47%; ₹4.09 lakh crore of net foreign financing. Worked Example 6: US real 1.7073%, India real 1.2019%, expected depreciation 1.4634%, and a dollar return from India of 3.7320%, which is 0.52 pp below the 4.25% available at home. Worked Example 3's deltas: PMI −2.7 against a −1.5 consensus miss, retail −2.1 against −1.0, CPI −0.7 against −0.2, curve +25 bp. All reproduced here. Then compute the thing Worked Example 4 states qualitatively and leaves as a watch-item. See the lens below, because the precise threshold is not the one the sentence implies.

Quantitative lens

A dashboard is a claim about the present, and three properties of that claim are computable: how old it is, how often it will cry wolf, and what its central conditional statement actually requires. All figures below come from the module's own stated cadences, indicator count and parameters; they are properties of that design, not measurements of any economy.

A "current" board is a mixture of vintages spanning five months. Take the release calendars §4.8 gives. At a randomly chosen moment, the expected age of the newest available observation is the publication lag plus half the frequency. That gives 0.5 days for the curve, the 10-year, the dollar, oil and the VIX; 10.5 days for FX reserves on the Friday Weekly Statistical Supplement; 16.0 days for GST collections and the PMI; 17.0 days for fortnightly bank credit; 27.0 days for CPI released around the 12th; 57.0 days for IIP with its six-week lag; 105.5 days for quarterly GDP; 135.5 days for the balance of payments; and 165.5 days for a quarterly-urban PLFS unemployment reading. The mean across those fourteen cells is 39.5 days, and the spread from the freshest cell to the stalest is 165 days, about 5.4 months. Two consequences follow and both are practical. First, the sentence "the economy is in a disinflationary slowdown right now" is, mechanically, a statement about a set of moments spread across half a year. Second, and this is the design instruction, the cells that move fastest are the financial ones, and the cells you most want for a structural read are the slowest. That is not a flaw to be fixed; it is why §4.9 tells you to weight the leading, market-priced indicators, and it is the reason the memo's regime call is a tilt rather than a nowcast. Put the release date in every cell and the board tells you this about itself.

An eighteen-cell board with a two-sigma trigger fires by accident in most months. §4.10 rightly demands falsification triggers, and the natural way to write one is per indicator: "this read breaks if CPI surprises high." Give that trigger a threshold in units of the series' own variability and the arithmetic is immediate. A two-sided two-sigma threshold fires 4.55% of the time for a single indicator under a no-change null. Across 18 indicators, the probability that at least one fires in a given month is 56.8%, with 0.82 expected firings. At one sigma it is 99.9% and 5.71 expected firings; even at two-and-a-half sigma it is 20.1%. So a board designed the obvious way will hand you a "your regime call is broken" alert in the majority of months when nothing has happened at all, and an analyst who responds to each one has replaced a standing instrument with a random-walk generator. Two fixes, both exact. Raise the per-indicator threshold to 2.757 sigma and the family-wise monthly false-fire rate across eighteen cells falls to 10%. Or, better, because it matches how regimes actually change, keep the two-sigma cells and require a cluster: the probability that three or more of eighteen breach two sigma by chance in the same month is 4.61%, against 19.64% for two or more and 56.75% for one or more. Three cells is the rule to write into the memo, and it has the additional virtue of being what a genuine regime shift looks like, since regime changes move growth, inflation and financial conditions together rather than one cell at a time.

And the transformation you choose is worth more than most of the surprises you are measuring. Take the CPI print from M7.03's Worked Example 1 and read the identical data three ways: year-on-year gives 4.0541%, three-month momentum compounded to an annual rate gives 3.1701%, and the same momentum multiplied by twelve gives 3.1250%. The spread between the year-on-year and the momentum read is 88 basis points on the same series in the same month, larger than most consensus misses anyone would call a surprise. This is why the transformation must be declared before the print, in writing, per series. An analyst who reaches for the year-on-year in a calm month and the momentum read at a turning point will always find the number that confirms the call, and will never notice doing it.

One number the module leaves as a direction, which deserves to be a threshold. Worked Example 4 concludes that "the entire resilience rests on nominal g staying above r," with r = 7.2%. That is right about the sign of the snowball and understates the requirement, because the country is also running a primary deficit that the snowball has to absorb. Set Δd = 0 in the exact law and solve for growth: g* = (r·d + p)/(d − p), which at d = 0.82, r = 7.2% and a primary deficit of 1.5% of GDP gives g* = 9.20%. So the debt ratio stabilises only above 9.20% nominal growth, not above 7.2%. The favourable snowball begins at r, but stabilisation needs the snowball to be large enough to swallow the deficit as well. Against the module's assumed g = 10.5% there is 130 basis points of headroom, which is a far more useful statement of the watch-item than "above r": it names the level, it is falsifiable against a single quarterly print, and it is exactly the kind of threshold a dashboard cell should carry. (This is M7.04's finding arriving from the India side: a favourable r − g gap enlarges the deficit you may run; it does not by itself stabilise anything.)

Honest limits, and for a capstone they are the deliverable as much as the dashboard is. The threshold arithmetic assumes the eighteen indicators are independent, and they emphatically are not. The curve, the ten-year and credit spreads move together; CPI and inflation expectations move together; the rupee and oil move together. Positive correlation reduces the family-wise false-fire rate below the 56.8% computed above and simultaneously makes the cluster rule easier to trip for a real reason and for a spurious one alike, so the true numbers sit somewhere below the independent case and the ordering of the conclusions survives while the magnitudes do not. The right fix is to estimate the correlation matrix from your own history and simulate, which is a genuine piece of work and worth doing once. The no-change null is also a fiction: macro series are autocorrelated and trending, so "two sigma of the monthly change" is not a clean significance statement, and the honest reading of these numbers is as an order-of-magnitude warning about board design rather than as a calibrated test. The staleness figures are expected ages under the module's own stated calendars, and real calendars slip, get revised, and differ between the first estimate and the settled one, which is the separate revision problem that only a vintage-aware store solves. The 9.20% stabilising growth rate is conditional on every input: d, r and the primary deficit are the module's illustrative figures, r is an effective average rate that moves slowly with the maturity profile rather than with the market, and the number moves when any of them does. And the largest limit of all is the one §4.10 already states and the arithmetic reinforces: none of this forecasts anything. A dashboard locates you; it does not tell you what happens next, the regime call is a tilt with triggers rather than a prediction, and the memo is graded on whether the read was defensible when it was written. Nothing on this page is a statement about the current repo rate, CPI, reserves, current account, fiscal deficit or growth rate. Every one of them is free, primary and dated at the sources §4.8 names, and the whole discipline of this capstone is that you go and get them yourself, with the release date written next to each.

Do it in code: hold the board as a specification (id, cadence, publication lag, transformation, threshold) rather than as a page of numbers; store vintages so a surprise can be reconstructed after the fact; put the release date in every cell and let the board display its own staleness; set thresholds in units of each series' variability and require a three-cell cluster before reopening the regime call; and write the memo's triggers as cell-level conditions the pipeline can evaluate, so next month's board tells you whether last month's thesis has been falsified rather than leaving you to remember. See G2/QD1.01, G2/QD2.04 and G1/QM1.01.

Where this goes next: galaxy cross-links

• G2 · QD1.01 Data Pipelines & Vendors. The fetch layer done properly: scheduling, retries, rate limits, vendor evaluation, and the vintage handling that makes a surprise reconstructible months later.
• G2 · QD2.04 Reproducible Research & Deployment. Turning the dashboard from a notebook into a standing instrument that runs on a schedule and produces the same output from the same inputs a year from now, which is what "updated monthly forever" actually requires.
• G2 · DA1.03 Pandas Cleaning & Time-Series Bridge. Resampling across daily, weekly, monthly and quarterly cells without silently mixing frequencies, and the alignment discipline the change-and-surprise lens depends on.
• G1 · QM1.01 Probability, Distributions & Estimation. The threshold and multiple-comparisons arithmetic above, done for real: correlated tests, family-wise error, and what a two-sigma move means in a series that is not independent month to month.
• G3 · CN1.03 The So-What. The regime memo is a so-what document; this is the discipline of forcing a defensible call, naming what would change it, and saying it in one page.

Teach it back & journal

Feynman prompt. Write a one-page explainer for a smart 15-year-old titled "Is India's economy actually as strong as everyone says?" You must convey, without jargon: that India grows fast mainly by selling to its own people (which makes it steady but slow to build factory jobs); that its scary-sounding government debt isn't scary the way Sri Lanka's was, because India owes the money in its own currency it can print (so the danger is higher prices, not going bust); that its biggest outside danger is the price of oil, because it buys almost all its oil from abroad; and the one honest catch, that having lots of young people only helps if they can get good jobs, which is the real question hanging over the whole story. If your page needs the words "current account," "inflation targeting," or "demographic dividend," you don't own it yet. Say it in plain words.

Journal prompt. Three entries, and then a habit. (1) Build the dashboard (the mini-project) and paste today's snapshot into your journal with the date. This is entry one of a monthly series you'll keep for years; the numbers you write today are the "as of" you'll compare against later. (2) Write your regime memo for this month and commit to its falsification triggers in writing. Then, at your next monthly review, come back and grade the read, not the outcome: was it defensible from the data? which trigger, if any, fired? (3) Reflect honestly: before this module, which did you believe: the boosterish "India is the next China, buy everything," or the declinist "jobless growth, it's all hype"? Write, in your own words, why both are analytically useless moods, and replace them with the one structural sentence you'd now defend about India's economy. This is the frame the module most wanted to install; writing it in your own words is how it sticks.

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