Seasonality, base effects and real-vs-nominal

The last module dealt with a number that was unfinished. This one deals with something trickier: a number that is completely accurate, fully revised, honestly produced — and still misleads you. Because a correct figure can tell a false story in three well-known ways, and all three fool careful people who would never be caught by a wrong number.

Inflation can "fall" while prices are still rising. A company's sales can "grow 12%" while it sells no more goods than last year. Factory output can "jump 30%" purely because one month is always busier than another. In each case nobody lied, no revision is pending, the arithmetic is exact — and the headline still points the wrong way. The three culprits are seasonality, base effects, and the gap between real and nominal. The question this module answers is: what is the single thing to check, in each case, to strip the distortion out and read what the number is actually telling you?

These three distortions deserve their own module because they are the ones that survive every other check. You can wait for the final revised figure, trust the statistics office completely, and read the number with full attention — and still be fooled, because the trick is not in the number's accuracy but in the comparison it is built on. A percentage change is always "this, versus that," and if the "that" is chosen or shaped by the calendar, by last year's oddity, or by inflation, the change describes the baseline as much as the present.

This is the deepest sense of the book's standing warning: a macro headline is a claim, and you strip the trick before you trust the number. In the previous module the trick was vintage — how finished the number is. Here the tricks are three different distortions of comparison, and each has a clean antidote. Learn the three antidotes and you gain a kind of X-ray vision: you can look at a headline everyone else is reacting to and see whether the move is real or a shadow cast by the baseline.

The stakes are practical. A reader who thinks inflation is genuinely falling, when it is only a base effect, will misjudge what the RBI is likely weighing. A reader who reads nominal sales growth as real expansion will overpay for a company that is merely passing on inflation. A reader who reads a seasonal jump as a recovery will call a turn that is just the festival season arriving on time. The distortions are ordinary; the errors they cause are expensive.

Take each in turn, with its antidote attached.

Seasonality. Many series have a regular within-year rhythm that repeats every year — seasonality. Factory output rises before festivals; farm output follows the harvest; many businesses push hard at quarter-end. Compare a naturally busy month to a naturally quiet one and you get a big change that is just the calendar repeating itself. The antidote: compare like with like — either use the seasonally-adjusted series, which mathematically removes the usual pattern, or compare the month to the same month last year (year-on-year), which faces the same season on both sides.

Base effects. A year-on-year change compares today to the same period a year ago — the base. If that base was unusual, the comparison is distorted regardless of what is happening now. A base effect is exactly this: a move in the year-on-year number caused by the oddness of the base period rather than by the present. If prices spiked a year ago, this year's inflation can fall on paper even while prices keep rising today, simply because the spike drops out of the twelve-month window. The antidote: look at what the base period was doing, and cross-check with the month-on-month change, which does not depend on the year-ago figure at all.

Real versus nominal. A number measured in rupees is nominal — it mixes together how much and at what price. When inflation is high, a rupee figure can rise sharply while the actual quantity of goods is flat, because each unit simply costs more. The real figure strips inflation out to show the change in true quantity. The antidote: subtract inflation. For a company, ask whether sales grew in units or only in rupees; for the economy, note that real GDP already removes inflation using a price index called the GDP deflator, while a nominal figure does not.

Three filters, three antidotes, one habit: never accept a percentage change until you know what it is being compared against — the season, the base year, and the price level. Miss any one filter and the shadow it casts reads as if it were the signal.

Watch a single inflation headline fool a careful reader, then get stripped clean. illustrative

The monthly CPI print lands and the headline reads: "Inflation drops sharply to 4% from 6%." The natural reading is relief — prices are coming under control, the cost of living is easing, perhaps the RBI has more room to cut. A reader takes the 4% as good news about today's prices.

Now apply the base filter. Inflation here is a year-on-year figure — this month's prices against the same month a year ago. Look at that base month: a year ago, food prices had spiked on a poor harvest, pushing the index unusually high. This year's index is being compared to that elevated base, so the twelve-month change looks small — even though, month-on-month, prices are still creeping up right now. The "drop to 4%" is largely last year's spike dropping out of the window, not today's prices falling. Cross-check with the month-on-month figure and you see prices still rising modestly. The relief was misread; the level of prices has not fallen at all, only the year-on-year rate of increase, and that mostly for base reasons.

The same trap wears a corporate suit. A composite consumer-goods maker reports sales up 12% in rupees, and the market cheers "strong growth." Apply the real-versus-nominal filter. Inflation ran high that year, and the company's unit volumes — packets sold — were flat. It simply raised prices in line with everyone else. Strip inflation out and real growth is close to zero: the same amount of soap, sold for more rupees. The 12% was the currency shrinking, dressed as the business expanding. A reader who paid up for "12% growth" bought inflation, not progress.

And the seasonal trap: factory output "jumps 30% from last month." But this month is festival-season production, which is always far higher than the quiet month before it. Compare it to the same month last year instead, and output is up a modest 3%. The 30% was the calendar, not a recovery.

Three accurate numbers, three false stories, three clean antidotes. In each case the fix was the same shape: find what the number is compared against, and correct for it.

Stripping out the three distortions gets you an honest read of the number. It does not get you everything, and it is worth being clear about the remaining limits.

A correctly stripped number still cannot predict the next one. Knowing that this month's inflation drop was a base effect tells you how to read this print honestly; it does not tell you where prices go next. The antidotes clean the past and present, not the future. Anyone converting "it was just a base effect" into a confident forecast is smuggling prediction into what was only a correction.

The distortions can also mask genuine change as well as manufacture false change. A base effect can hide a real acceleration as easily as invent a fake slowdown; seasonality can bury a true improvement inside an ordinary busy month. So the antidotes are not "assume every move is an illusion" — they are "separate the real move from the distortion," which sometimes reveals that the move was real after all. The skill is subtraction, not blanket suspicion.

And a clean macro number still has to be transmitted to a company through all the steps this book has traced. "Real growth was modest, once you strip inflation" is an honest read of the economy; what it means for a particular lender or maker runs through the transmission chain, not straight from the headline.

These are the distortions that catch the most careful readers, precisely because the numbers are correct.

1. Reading a base-driven inflation fall as prices falling. A lower year-on-year rate can sit on top of prices that are still rising today. Always check the base month and the month-on-month move before you feel relief.

2. Mistaking nominal growth for real growth. A rupee figure includes inflation. Sales "up 12%" with flat volumes in a high-inflation year is the currency shrinking, not the business growing. Ask about units, not just rupees.

3. Reading a seasonal swing as a trend change. A big month-on-month jump can be the festival season or the harvest arriving on schedule. Compare like month with like — year-on-year or seasonally adjusted — before calling a recovery or a slump.

4. Comparing across two different distortions at once. A number can carry a base effect and be nominal and be seasonal all together. Strip all three; correcting one while ignoring the others still leaves you fooled.

5. Turning an antidote into a forecast. "It was just a base effect" honestly explains the present print; it does not predict the next. Using the correction as a prediction re-introduces exactly the false confidence the correction removed.

Underneath the three antidotes sits a single question, and it is worth carrying as the take-away of this whole module: compared against what? Every percentage change is a comparison, and the comparison is where the distortion hides. Compared against a naturally quieter season? That is seasonality. Compared against an unusual base period? That is a base effect. Compared in rupees while the rupee itself is losing value? That is nominal masking real. Ask "compared against what?" and you are led straight to whichever of the three filters the number needs.

This is the honest reader's edge over the headline reader, and it costs nothing but a pause. While others react to "inflation falls," "sales surge," "output jumps," you ask what each is measured against, apply the one or two filters that fit, and read the true underlying move — which is often smaller, sometimes larger, and occasionally the opposite of what the headline implied. You will not predict the next number any better; that was never the promise. But you will stop being fooled by the shadow the baseline casts, which is one of the most common and most expensive ways careful people misread the economy. Combined with the previous module's discipline on vintage, you now have the two halves of reading data honestly: how finished a number is, and what it is compared against.