What macro cannot tell you

This module closes the first part of the book, and it does so deliberately by gathering all the ways macro fails into one honest inventory. Not to make you cynical — the rest of the book is built on reading macro carefully — but to fix the boundaries of the tool before you pick it up in earnest. A power tool used within its limits is invaluable; the same tool used past its limits takes your fingers off.

So far Part One has built up a method: read macro as weather (module one), trace it through the six channels (module two), never confuse the economy with the market (module three), and position rather than trade (module four). This module is the counterweight to all of it — the standing list of what a macro number can never tell you, no matter how carefully you read it.

Keep this list close. Every later module in the book, on rates or the rupee or crude or the Budget, is subject to it. The number you are handed is always late, often revised, sometimes distorted by last year's arithmetic, frequently already in the price, and occasionally busy changing the very thing it claims to measure. Reading it well begins with knowing all of that.

The deepest trap in macro is treating the number as reality itself. It is not. A macro number is a measurement — a human-made estimate, built with assumptions, of something too vast to observe directly. And measurements can be late, wrong, distorted, and misleading even when honestly made.

Start with the most basic point, the one the statistician Darrell Huff spent a whole book on: a statistic is an argument in disguise, and it deserves to be cross-examined, not believed. How was it collected? What does it leave out? Against what is it compared? A growth figure, an inflation print, an employment number — each is the end of a long chain of choices about what to count and how. That does not make them useless; it makes them evidence, to be weighed, not verdicts to be obeyed.

Then add the specific defects, because they recur in every macro series you will read:

Lag. A lag is the delay between something happening and its effect showing up — in the economy or in the data. A rate change takes many months to reach borrowing and demand; a growth figure describes a quarter already gone. You are always reading the past and waiting on the future.

Revision. A first estimate is a draft. Data revision — the later restatement of a figure as fuller information arrives — can move a number meaningfully, months after everyone reacted to the first version. Act on the draft as if it were final and a revision can whipsaw you.

Base effects. A base effect is a swing in a year-on-year number caused mainly by an unusual level a year earlier, not by anything new now. If last year's month was freakishly high, this year's comparison looks low even if prices are flat today. The number moved because of old arithmetic, not current reality.

Reflexivity. This is the strangest and most important. The act of measuring and predicting the economy can change the economy itself — George Soros's word for the feedback loop between perception and reality. A widely believed forecast of a boom can cause the borrowing and spending that make the boom, until the belief and the reality feed each other past the point of sense. The macro variable is not a fixed target you shoot at; it watches you aim and moves.

And there is a quieter danger sitting on top of all these. When macro looks calmest and most benign — steady growth, low inflation, easy money — is often when the most risk is being built up beneath the surface, because calm encourages leverage and complacency. A long stretch of stability quietly breeds the excess that ends it, as Hyman Minsky argued. A reassuring macro read can be most dangerous precisely when it is most reassuring.

Put the defects together as a gauntlet. Before any macro number becomes something you act on, it should survive five filters. What comes out the far end is not a forecast or a signal — it is a modest, humble read, hedged with everything the filters removed.

Walk the five once, because you will run them on every number in this book.

Filter one — is it provisional? Ask whether this is an early estimate that will be revised. If so, hold it loosely. A first growth print is a pencil sketch, not a photograph.

Filter two — is there a base effect? Ask what last year's comparison level was. A year-on-year number that looks dramatic may be dramatic only because of an odd base a year ago. Cross-check with the shorter-term trend to see current momentum.

Filter three — is there a lag? Ask whether the effect has even arrived yet. A policy change or a shock may take many months to reach the accounts. What you are reading may be the delayed echo of an old event, or a new event whose effect is still months away.

Filter four — is it already priced? Ask whether the market has already assumed this. A public number everyone can see is the crowd's shared knowledge, and the crowd has usually already acted. Your read pays only where it differs from what is already in the price.

Filter five — is it reflexive? Ask whether the belief is changing the thing itself. A widely held macro view can create the behaviour that confirms it — until it doesn't. The most consensual, comfortable read is often the one quietly building the most risk.

Take the most instructive failure in macro investing — being right and making nothing. illustrative

An investor forms a well-reasoned view: inflation is likely to ease over the coming year, which should eventually let the RBI cut rates, which should help rate-sensitive sectors. Trace it through the method and it is a clean chain — the financing channel, rate-sensitive sectors, the interest-cost line. And, as it happens, the view proves broadly correct: inflation does cool and rates do come down.

Yet the position she takes makes almost nothing. Why?

Run the filters. By the time she acted, the easing was already widely expected — economists had forecast it, commentators had discussed it, and the rate-sensitive names had already risen months earlier in anticipation (filter four: priced in). Part of the apparent cooling in inflation was a base effect — last year's spike falling out of the year-on-year comparison — which everyone could see coming, so it surprised no one (filter two). And the actual rate cut, when it came, arrived with a long lag and in line with expectations, so it moved prices barely at all (filter three).

Her chain was right. Her direction was right. She still made nothing, because every part of her correct view was public, expected, and already in the price before she acted. This is the humbling heart of macro: being right is not the same as being paid. You are paid only for the gap between what happens and what was already assumed — and a well-known macro view has almost no gap left in it.

This whole module is the "what it cannot tell you" section for Part One, so here is the inventory in one place, to carry forward.

Macro cannot tell you what to buy. It names conditions over an economy, never a company. The bridge from condition to company is the transmission chain plus a careful reading of the specific business — and often that work reverses the naive conclusion.

Macro cannot tell you the timing. Lags are long and variable, the data is late, and the cycle turns when it turns. Even a correct direction can arrive years off schedule, and early behaves exactly like wrong. A confident date attached to a macro turn is fortune-telling with a spreadsheet.

Macro cannot tell you what is already priced. The value in any public number is only the part the crowd had wrong, and a number everyone can see is, almost by definition, already reflected. Correct-but-consensus pays nothing.

Macro cannot give you a clean, final figure. First prints are drafts, revisions move them, and base effects distort the year-on-year story. The number is evidence to weigh, never a fact to obey.

And macro cannot promise that a calm reading means safety. The most benign backdrop can be the one quietly building the most risk, because calm invites leverage and complacency. A comfortable macro read deserves more suspicion, not less.

These are the errors this whole module exists to inoculate against. Meet them once here; you will recognise them everywhere in the book.

1. Treating the first print as final. Acting hard on an early estimate that will likely be revised. The draft is not the photograph.

2. Believing a base-effect swing. Reading a year-on-year drop or jump as fresh news when it is mostly last year's arithmetic falling out of the comparison.

3. Ignoring the lag. Expecting a policy change or shock to hit immediately, then being surprised when the effect arrives months later — or acting as if a still-distant effect is already here.

4. Confusing being right with being paid. Forgetting that a correct, widely shared macro view is already in the price, so it earns nothing. The edge is in the gap from consensus, not in the direction.

5. Trusting the calm. Reading a benign, stable macro picture as safety, when calm is often exactly what breeds the leverage and excess that end it.