Your shop friend says: "Relax, I'll mark the kurta up by 20\% for the festival, and after Diwali I'll mark it down by the same 20\%. The price comes right back to where it started." It sounds perfectly fair. It is not. The kurta that started at ₹1000 finishes at ₹960 — a quiet 4\% below where it began. The two 20s do not cancel, and the tower below shows you exactly why.
The tower
Why the two twenties are not equal
The trap hides in one small word: "20\%." A percentage is never a standalone rupee amount. It is a rupee amount times a base. And the base silently changes between the two steps.
- Step up. 20\% of the starting price ₹1000 is ₹200. New price: ₹1200.
- Step down. 20\% of the new price ₹1200 is ₹240. New price: ₹1200 - 240 = 960.
The first 20\% was computed from ₹1000. The second 20\% was computed from ₹1200. They were never the same rupee amount — one was ₹200, the other ₹240. Subtracting a bigger number than you added is exactly why you finish below the start.
Why: every percentage operation is a multiplication, and "of what" matters. "20\% of 1000" and "20\% of 1200" are two different numbers, even though both are "20\%". The percentage sign does not carry the base around with it — the base has to be supplied freshly each time.
The multiplier form
There is a cleaner way to see this whole dance. Every percentage change can be written as a single multiplier.
- A 20\% increase multiplies by 1 + \tfrac{20}{100} = 1.20.
- A 20\% decrease multiplies by 1 - \tfrac{20}{100} = 0.80.
Chaining the two steps means multiplying by both:
So the composite multiplier is 0.96, not 1. The "4\% loss" you see on the tower is the gap between 0.96 and 1. And this is not a rounding error — it is exact algebra:
Plug in p = 0.20 and you get 1 - 0.04 = 0.96. The tail-end of the formula, -p^2, is the shortfall. For any percentage p (as a decimal), marking up by p and then marking down by p leaves you at 1 - p^2 of the original. You always end below.
The symmetry direction doesn't help
What if the shop friend had done it in the opposite order — 20\% off first, then 20\% up?
Same answer. Multiplication is commutative, so swapping the order of a markdown and a markup gives the same final price. The two 20s cannot cancel no matter which one you fire first.
Example: +50% then −50% on a ₹$1000$ shirt
Start at ₹1000. Apply +50\% first: the shirt jumps to ₹1500 (you added ₹500). Now apply -50\%: you take half off ₹1500, which is ₹750 — and you finish at \textsf{₹}750, a full \textsf{₹}250 below where you started.
Check with the formula: 1000 \times 1.5 \times 0.5 = 1000 \times 0.75 = 750. Or directly 1 - 0.5^2 = 0.75. The bigger the percentage, the bigger the shortfall — because p^2 grows fast. A 50\% swing leaves a 25\% gap. A 10\% swing leaves only a 1\% gap.
The common confusion, named
"I added 20 and subtracted 20, so the two cancel." This reasoning is correct for rupee amounts but wrong for percentages. If the shop friend had written "I'll add ₹200 and then take off ₹200" the kurta really would return to ₹1000. The word "20\%" is what breaks the symmetry, because the rupee amount it stands for depends on what it is a percentage of.
This is also why a shopkeeper's "buy two, get 50\% off" often beats "50\% off everything, always" — the percentage alone does not tell you how much money changes hands. The base does.
Summary
- A 20\% markup followed by a 20\% markdown does not return to the start; it lands at 96\% of the original.
- The reason: the markdown is taken off a larger base (the price after the markup), so the rupees removed exceed the rupees added.
- The algebra is (1 + p)(1 - p) = 1 - p^2. The -p^2 is the shortfall, and it is always negative — you always end below.
- Bigger percentages amplify the gap: p = 0.50 loses 25\%, p = 0.20 loses 4\%, p = 0.10 loses 1\%.
- Order doesn't matter. (+p) then (-p) or (-p) then (+p) give the same final price.
Related: Percentages and Ratios · 50% Off Then 30% Off Is Not 80% Off · Why You Can't Add Successive Discount Percentages · Does a 200% Increase Mean Doubling?