Price elasticity of demand calculator (point and arc)
The till rarely moves one-for-one with the ticket. Cut a price by 10% and volume might rise 3%, or 40%. The price elasticity of demand (PED) puts a number on that sensitivity: the percentage change in quantity sold for a one-percent change in price. It is the calculation that precedes a tag change, and one of the first in a microeconomics course.
The coefficient is almost always negative: price up, quantity down. What gets quoted is the absolute value. Above 1, demand is elastic and a price rise cuts revenue. Below 1, demand is inelastic and the rise adds to the till.
Two formulae sit side by side, and they do not agree. Point elasticity divides by the starting values; arc elasticity divides by the midpoints, which makes it symmetric. Both are printed below. Once the move runs past a few percent, the arc figure is the one to keep.
Two price-quantity pairs
Starred jargon is unpacked in the glossary at the foot of the article.
The revenue test: what |PED| says about a price move
Revenue is price times quantity. Nudge the tag and quantity moves the other way. Which of those two wins is an elasticity question, not a hunch. The revenue test* is the practical reading of |PED|*, and it fits on three lines.
- If |e| > 1, demand is elastic: a price cut raises revenue, because the extra units more than cover the thinner ticket. A rise does the opposite.
- If |e| < 1, demand is inelastic: a price rise raises revenue. Volume shrinks, but not by enough to offset the extra pounds per unit.
- If |e| = 1, revenue is already at its peak. Move either way and the till takes less.
The rule is local. It describes a small step around the point measured, not a leap from £8 to £18 on a curve never seen. In practice the coefficient is re-estimated after each meaningful change rather than ridden for a year.
TR = P × Q · revenue maximised where |e| = 1
Why textbooks drop the minus sign
Price up, quantity down. That is the law of demand, and it is why the elasticity calculated here is almost always a negative number. Raise the ticket from £10 to £12 and seats do not fill. They empty. The coefficient records that: a minus sign, by construction, not as a verdict on the product.
Nobody in a pricing meeting says “minus two.” They say “two.” The figure that gets quoted is the absolute value |e|. Larger |e| means a sharper response. An elasticity of −2 is called more elastic than −0.5, even though −2 is the smaller algebraic number. The minus only tells the direction, which was already known. |e| tells the size, which is the bit that changes a decision.
A positive coefficient does show up now and then. A Veblen good* that uses price as a quality signal, a Giffen staple*, or — far more often — two months of data where something else moved at the same time (a campaign, a rival’s collapse, a heatwave). Treat a plus sign as a warning to check the experiment, not as a new law of demand.
Arc elasticity versus point elasticity
Same two prices, two different answers. The argument is which base to divide by.
Point elasticity* takes the starting pair as the whole world: ΔQ against Q1, ΔP against P1. Fine when the move is tiny — a penny on a pound, a week of data. Push the change to 20% and the starting point is a poor description of the journey. Go from £10 to £12 and one coefficient appears; go from £12 back to £10 and another appears. The market did not change. The arithmetic did.
Arc elasticity* (the midpoint method, sometimes called Allen’s) divides by the average of the two prices and the average of the two quantities. The trip A→B and the trip B→A then give the same number, which is what two observed points require. That said, neither formula is “the” elasticity of the good. Both describe a particular move on a particular curve.
e = (ΔQ / Q1) ÷ (ΔP / P1) · point elasticity, initial base
e = (ΔQ / ((Q1+Q2)/2)) ÷ (ΔP / ((P1+P2)/2)) · arc elasticity
For a tag change of more than a few percent, trust the arc figure. This calculator prints both so the drift between them is visible.
Five labels, from a vertical demand curve to a horizontal one
Once |e| is in hand, the textbook splits demand into five buckets. They are labels, not laws of nature, but they save a lot of fumbling in a meeting.
- Perfectly inelastic (|e| = 0) — quantity does not budge. Insulin for a dependent patient is the usual picture: the dose is the dose.
- Inelastic (0 < |e| < 1) — quantity moves, but by less than price. Bread, milk, the commute. A 10% rise costs maybe 3% of volume.
- Unit elastic* (|e| = 1) — percentage in, percentage out, revenue sits still. On a straight demand line this is the midpoint, and it is where revenue peaks.
- Elastic (1 < |e| < ∞) — quantity jumps more than price. Leisure flights, a branded jumper with three lookalikes on the next rail.
- Perfectly elastic (|e| = ∞) — the smallest rise and quantity demanded drops to zero. That is the demand facing one price-taking firm in perfect competition, not market demand, which still slopes down even then.
Real products live in the middle three. The two extremes are teaching tools — and a reminder that “inelastic” never meant “a brick.”
Profit sits one step past revenue
Maximum revenue is not maximum profit. Elasticity speaks only to the till, never to the cost of the extra unit.
A streaming service that fills up at £7 a month can post a handsome revenue number and still lose money if each extra subscriber costs £4 in licences, bandwidth and support. A café that puts 40p on a flat white may sell a few fewer cups and make more, because the milk and the barista were already there. The missing ingredients are marginal cost* and the competitive set: a rival one click away, a contract that cannot be walked away from, a brand not worth cheapening.
Use the coefficient to sketch the revenue side. Then put cost next to it. The price that survives is the one that leaves the most after both.
What keeps demand inelastic
Why does one product shrug off a rise while another collapses? Four regulars, and they stack.
- Necessity — water, electricity, a life-saving drug. The dose is not postponed because the pack went up 8%.
- No close substitute — if the alternative is a different product, or a long journey, buyers swallow the price. A unique spare part behaves this way; a supermarket own-brand baked bean does not.
- Tiny budget share — table salt, a biro, a box of matches. Even a doubling barely registers on the month’s spend, so nobody shops around.
- Habit and addiction — tobacco, a daily coffee order, a brand bought for twenty years. Behavioural inertia is a real cost of switching, and it shows up as a smaller |e|.
Flip any of those and demand loosens. A product with three rivals, a fat line on the bank statement, and no habit behind it will be elastic before the sentence is finished.
Horizon: short run and long run
Give people a week and they grumble. Give them three years and they change the car, the boiler, the contract.
Short-run demand is rigid because the costly adjustments take time: a different commute, a smaller flat, a cancelled gym, a heat pump. Long-run demand is more elastic for the same good. Petrol is the textbook case — estimates sit around −0.1 to −0.3 in the short run, and can reach −0.6 to −0.8 once drivers have had time to buy a more efficient car, move closer to work, or switch to the train. Household energy follows the same pattern. A streaming subscription is stickier month-to-month than it looks on a two-year view, once the free trial has ended and the password is shared with half the household.
When an elasticity is quoted, the horizon has to be named. A Chancellor raising fuel duty and a product manager testing next month’s tariff are not looking at the same number.
Ballpark numbers from applied work
Published estimates wander by country and by decade. The ranges below are absolute values, the ones people actually quote, and they are orders of magnitude — not a forecast for a given SKU.
- Staple foods — 0.2 to 0.5 (inelastic). People still eat.
- Tobacco — 0.3 to 0.6 (inelastic, with a generational slide as younger cohorts smoke less).
- Petrol — 0.1 to 0.3 short run, 0.6 to 0.8 long run.
- Air travel, leisure — 1.5 to 2.5 (elastic). Business travel sits lower; the Friday-night city break does not.
- Luxury, high-end cars — 2 to 4. Plenty of substitutes, and the purchase can wait.
The market definition does half the work. “Food” is inelastic; “this brand of cheddar in this shop this week” is not. Narrow the box and |e| jumps.
Income, cross-price, and supply elasticities
Price elasticity of demand is one cousin in a family. The others answer different questions, and mixing them up is a classic exam — and board-paper — mess.
- Income elasticity of demand — % change in quantity for a 1% change in income. Positive for a normal good, negative for an inferior one. Above 1, people call it a luxury: income up 10%, sales up more than 10%.
- Cross-price elasticity — % change in quantity of A when the price of B moves. Positive if they are substitutes (butter and margarine), negative if they are complements (printers and ink, cars and petrol).
- Price elasticity of supply — how much quantity supplied moves when price moves. Generally positive. A factory that can add a shift overnight is elastic; a vineyard that needs five years is not.
A supermarket that only watches own-price elasticity will miss the rival’s 15% off, which is a cross-price story, and a recession, which is an income story. Own-price, supply and the crossing of the two curves are what set a market price in the first place.
Tax incidence and the inelastic side of the market
Put 20p of duty on a litre and the pump price does not always rise by 20p. Who eats the tax — the buyer or the seller — is an elasticity question, not a legal one. The statute names the firm. The market names the wallet. That sharing is tax incidence*.
If demand is very inelastic and supply is slack, buyers take most of it: they were not going to cut back much to begin with. If demand is elastic and supply is stuck (a harvest already in, a plane already fuelled), the seller absorbs more, because raising the ticket would lose too many units. The rule of thumb is blunt and useful: the less elastic side of the market bears more of the burden.
A duty shifts a curve. The new crossing says who pays — still the same supply-and-demand intersection, read after the shift. “Passed on in full” is a claim about relative slopes, not a moral fact. A 20% VAT line on the receipt does not tell that the pre-tax price would have been 20% lower.
Estimating PED: log-log and identification
Two points from an A/B test are the clean case. Scanner data, a year of list prices, a panel of shops — that is messier, and the workhorse is a log-log regression*:
ln(Q) = α + β·ln(P) + γ·X + ε
β is the constant price elasticity (Cobb–Douglas form)
The coefficient on ln(P) is then the elasticity, the same at every point. Convenient. Also fragile.
- Endogeneity* — price moves with things not put in the regression: quality, brand, season, a TV advert. Ordinary least squares then blames the tag for sales that the advert caused, and β is biased.
- Simultaneity — supply and demand set P and Q together. An instrument is needed (a cost shock, a tax change, a weather event on the harvest) that shifts one curve and not the other.
- The window and the grain matter. Weekly store data is not annual national data. A loyal club-card segment is not the passing trade. Average them and the customers who actually switch disappear.
Even so, a well-designed price test beats a clever regression on observational data. Randomise the tag, keep everything else still, and the two-point calculation on this page is exactly what is needed.
A subscription price test, worked through
A UK streaming service sits at £10 a month with 100,000 subscribers — £1,000,000 of monthly revenue. It wants a better price. It does not guess. It runs an A/B test on a sample of new visitors:
- Group A, £9: sign-up rate 12%
- Group B, £11: sign-up rate 8%
Arc elasticity from those two cells:
- ΔQ (%) = (8 − 12) / ((8 + 12) / 2) = −4 / 10 = −40%
- ΔP (%) = (11 − 9) / ((11 + 9) / 2) = 2 / 10 = +20%
- |earc| = 40 / 20 = 2.0 — elastic demand
Because |e| > 1, a cut should lift revenue. They try £9.50 on the whole book — a 5% cut from £10. With |earc| = 2.0, that points to roughly a 10% rise in subscribers, to 110,000. Revenue goes from £1,000,000 to £1,045,000 (110,000 × 9.50). After all, the coefficient did its job on the till. What it cannot do is count the extra bandwidth bill, the support tickets, the licence that prices per stream. If those variable costs eat the £45,000, the “better” price is a worse business. The price that survives is the one that maximises profit, given costs and given what a rival will do the week after. One test is a snapshot. Run it again after the Christmas campaigns, and after a competitor launches at £5.99 with adverts. Elasticity moves.
99p endings and a kink in the curve
Crossing £10 is not the same as sliding from £9.40 to £9.70. Demand is not a smooth line. It has steps — “round” prices beyond which a slice of buyers drops out at once. An |e| measured between £9.99 and £10.20 will look larger than the same twenty-one pence between £9.40 and £9.61.
This calculator joins two points with a straight line. Between those two points, that is fair. If a psychological threshold sits in the gap, the slope is a kink, not a chord. Worth noting before a catalogue is rounded “because £10 is easier to say.”
Advance fares: one seat, several tickets
Leisure air travel sits around 1.5 to 2.5. A Tuesday-morning business trip with a fixed slot is far more inelastic. That is why GWR, LNER and the airlines sell the same seat at three or four fares: advance, off-peak, anytime, first. The cheap ticket skims the elastic weekend and the student. The expensive one keeps the captive Tuesday.
It is not a trick. It is the revenue test applied to two clienteles that barely mix. Price everyone like an advance fare and the trains fill while the inflexible revenue disappears. Price everyone like an anytime ticket and Saturday empties. A single pair (P, Q) yields the average elasticity of the market. A carrier has every reason to split it.
Fuel duty: miles that barely twitch
A 10p rise on a litre of petrol is front-page news. Miles driven that week barely twitch. Short-run fuel demand in Britain is stubborn for the reasons above: the car is already on the drive, the school run is not optional, and a month is too short to buy a smaller engine. That is the −0.1 to −0.3 band in UK clothing.
Fuel duty plus VAT is a large slice of the pump price, which is why Chancellors treat a duty freeze as a cost-of-living gesture and a duty rise as a revenue raiser that will not empty the roads overnight. Give it years and the picture softens: more efficient cars, more working from home, more rail where it exists. Long-run |e| nearer 0.6 to 0.8 is the version the Department for Transport has to live with when it models a decade of duty. The same litre is a different product on a Friday night and in a 2030 fleet forecast. Quote the horizon, or the number is theatre.
Glossary
- Revenue test
- The rule that links |PED| to the till: cut the price when demand is elastic, raise it when demand is inelastic, leave it when |e| = 1. It maximises revenue, not profit.
- PED
- Price elasticity of demand: the percentage change in quantity demanded for a 1% change in price, other things equal. Almost always negative; quoted as an absolute value.
- Veblen good
- A prestige item whose demand can rise with its price, because the tag itself signals quality or status. PED then comes out positive, against the usual law of demand.
- Giffen good
- A staple, for a very constrained household, whose price rise can paradoxically raise the quantity consumed, because still-dearer alternatives are cut even harder.
- Point elasticity
- The ratio of percentage changes calculated on the starting values (Q1, P1). Simple, but asymmetric: A→B does not match B→A.
- Arc elasticity
- The same ratio, calculated on the averages of the two prices and the two quantities (midpoint, or Allen, method). Symmetric, and the safer reading once the price jump is more than tiny.
- Unit elastic
- The case |e| = 1: quantity moves by the same percentage as price, in the opposite direction. Total revenue is then at its peak along the curve.
- Marginal cost
- The cost of producing one extra unit. Elasticity is silent on this line: it speaks to revenue, not to profit.
- Tax incidence
- The real split of a tax between buyer and seller once prices have adjusted. Not the name on the receipt: the less elastic side of the market carries more of the load.
- Log-log regression
- A regression of ln(Q) on ln(P). The coefficient on ln(P) reads directly as an elasticity, constant along the curve (a Cobb–Douglas form).
- Endogeneity
- Price is not set “from outside”: it moves with quality, season, brand — things that also move sales. A naïve regression then credits the tag with part of those other effects.