Potential GDP and the Output Gap: Why Capacity Is Estimated

Potential output, also called potential gross domestic product (GDP), is what an economy could produce with the workers, capital and know how it already has, without straining. The output gap is the distance between what an economy actually produced and that level, written as a percentage of potential. A negative gap means capacity is sitting idle. Potential is never measured, only estimated, and every estimate carries a judgement about what capacity even is.

An output gap worked out from two levels

Read the default setting as ordinary text. Sankhya, invented for teaching, produced Rs 17,47,200 crore in year 3. Set an assumed capacity path growing 6.00 per cent a year from Rs 16,00,000 crore in year 1 and potential for year 3 works out at Rs 17,97,760 crore, so the difference is Rs 50,560 crore below potential and the gap is minus 2.81 per cent. Year 2 on the same assumption gives minus 0.94 per cent. The shortfall widened by 1.87 points across a year in which output itself grew 4.00 per cent. Now change the one number nobody counted. Take the capacity path at 4.50 per cent instead and year 3 potential becomes Rs 17,47,240 crore, the difference falls to Rs 40 crore, and the gap rounds to nothing at all. Not one rupee of counted output moved between those two readings, and the gap went from a visible shortfall to nil.

Everything in that answer rests on one asymmetry. Actual output is a count. Somebody adds up what was produced, gets it partly wrong, revises it, and publishes a number that at least refers to events that happened. Potential output refers to events that did not happen: a version of the same year in which nothing was idle and nothing was strained. Nobody has ever seen that year. So an output gap subtracts a counted number from an imagined one, and the imagined one does all the arguing.

What is potential output, and what is the word straining doing in that definition?

Potential output is the level of production an economy can hold without strain. Not the level it can touch for one quarter by running plants through the night, deferring maintenance, and paying for overtime that nobody would agree to twice. The everyday version is a household kitchen. A household that normally cooks for six can cook for thirty at a wedding. The household has done it before and will do it again. Cooking for thirty every evening is beyond it. The cook does not sleep, the gas cylinder runs out on a schedule nobody planned for, and the vessels do not get washed. Thirty is the maximum and six is the potential, and writing thirty into the capacity estimate would make a household cooking for six every night look like a catastrophic shortfall.

The word straining is doing all of the work in the definition, and it is the word most often dropped. Drop the word and potential becomes a ceiling. Everything in the world sits below its ceiling all of the time, so a ceiling is a useless reference point. An economy running at its sustainable level is normal. An economy running above it is doing something it cannot keep doing. The sustainable level matters for that reason alone: it separates output that can be planned around from output that is borrowed from next year.

What actually sets an economy's potential output?

Three things, and every one of them moves slowly. The first is how many people are available to work. The size of that group runs on how many people there are, how many of them are of working age, and how many of those are willing to look for work. The second is the stock of things they have to work with: the machines, the buildings, the roads, the power lines, the software. The third is how effectively the first two are combined. Effectiveness covers skill, organisation, and whether a worker with a machine is put where the machine is useful.

Not one of those three can turn inside a quarter, and that single fact is why potential output is drawn as a smooth line while actual output is drawn as a jumpy one. A generation of workers arrives over two decades. A power plant takes years to build and then stands for thirty. Skill accumulates through schooling and through doing the work. Meanwhile orders dry up, a monsoon fails, or a port is blocked, and actual output falls or jumps within three months. A chart with one line that wanders and one line that glides is not showing a drawing convention. The chart is showing the difference between things that change fast and things that do not.

The capital stockThe accumulated pile of machines, buildings, vehicles, power lines and other durable equipment an economy has built up and not yet worn out. A stock changes only by what gets added or scrapped in a year. is the clearest case. Building factories takes time and takes the same workers and machines, so Sankhya cannot double its factories in a year even if every rupee of its output were spent on building them. The limit is one of physics and construction schedules, not of will.

How is the output gap computed, and what belongs in the denominator?

Actual output less potential output, divided by potential output and multiplied by a hundred, gives the gap in per cent. Sankhya year 3 gives actual output of Rs 17,47,200 crore against potential of Rs 17,97,760 crore. The shortfall is Rs 50,560 crore. The shortfall over potential is a quotient of 0.028124, so the gap reads minus 2.81 per cent once rounded. Year 2 runs the same way, with Rs 16,80,000 crore standing against Rs 16,96,000 crore for a shortfall of Rs 16,000 crore, and that shortfall over potential gives minus 0.94 per cent.

The denominator is where a quiet error lives. Both figures are sitting there, both are in Rs crore, both are roughly seventeen and a half lakh crore, and it feels as though it should not matter which one is divided by. It does. Rs 50,560 crore over Rs 17,47,200 crore, the actual figure, gives minus 2.89 per cent instead of minus 2.81. Actual sits on the wrong side of potential in each case, so dividing by actual output overstates a negative gap and understates a positive one. The error is silent. Both answers look entirely reasonable. Eight hundredths of a point is small here. The error is not small when the gap is large, and never small when two people are comparing two gap figures and cannot work out why they differ.

How can an economy grow and fall further behind at the same time?

Sankhya grew in year 2 and grew again in year 3. Real output stood at Rs 16,00,000 crore, climbed 5.00 per cent to reach Rs 16,80,000 crore, then climbed 4.00 per cent more to reach Rs 17,47,200 crore. Over the same two years the gap against potential went from nil, to minus 0.94 per cent, to minus 2.81 per cent. Output rose in both years and the shortfall grew in both years. Growth alongside a widening shortfall is not a contradiction and not a mistake in either series. A gap is a comparison against a line that is itself moving, so the direction of the gap is decided by which line moved faster and not by whether output rose.

Here it is in one arithmetic sentence. Potential grew 6.00 per cent in each year and actual output grew 5.00 and then 4.00. Over two years potential is up by a factor of 1.06 times 1.06, or 1.1236, and actual is up by 1.05 times 1.04, or 1.0920. The second divided by the first is 0.971876, a shortfall of 2.81 per cent. CompoundingApplying a growth rate to a base that already includes the previous years growth, so the effects multiply rather than add. Two years at 6.00 per cent gives 12.36 per cent in total, not 12.00. is why the gap does not simply add up: the two rates multiply and the divergence between them accumulates.

A household version helps. A cousin who studies four hours a night is improving. If the syllabus is expanding by six chapters a week and she is covering five, she is learning more every week and falling further behind the course at the same time. Both sentences are true. Insisting that improving and falling behind cannot happen together assumes a syllabus standing still, and this syllabus is not.

What does a negative output gap actually mean, and what does it not mean?

A negative output gap says one thing: some capacity that could have been used was not used. People who would have taken work did not have work. Machines that could have run stood still. Buildings that could have been full were not full. Sankhya year 3 with a gap of minus 2.81 per cent is an economy in which roughly Rs 50,560 crore of production that the estimated capacity could have carried did not happen, on the estimate being used. The number claims that much, and the claim stops there.

The list of things a gap does not say is longer and more useful. A gap describes a year that is over, so it is not a forecast. A shortfall against capacity has many possible causes and the number distinguishes none of them, so a gap is not a verdict on anybody. A gap does not say what anyone should do next. What to do is a separate subject, and no measurement settles it. One of the two numbers going into the subtraction was never observed, so a gap is not a measured quantity in the ordinary sense. A published gap of minus 2.81 per cent is a difference between something counted and something assumed, and it inherits every weakness of the second.

The word people reach for here is slackSpare capacity that is available but idle. In an economy it means workers who could be employed and equipment that could be running but is not, so output could rise without anything new being built., and it is a good word as long as it is held loosely. Slack is a statement about the size of the room between where an economy is and where it could sit without strain. Slack is not a measurement of anybody, not a claim about next year, and not a scorecard.

Why is potential output estimated rather than measured?

Because there is nothing there to measure. Actual output is a count of things that were produced, and although the count is difficult and gets revised, the events being counted happened. Potential output is a statement about a version of the year in which nothing was idle and nothing was strained, and that version of the year never took place. Potential output is a counterfactualA statement about what would have happened under conditions that did not hold. A counterfactual cannot be observed directly, only worked out from a model of how the thing behaves, so two reasonable models can give two different answers., and a counterfactual cannot be observed by anybody, however good their data collection is.

So potential has to be worked out from something else, and every method for doing that embeds a judgement. One approach fits a smooth trendA smooth line drawn through a jumpy series to represent its underlying direction, with the ups and downs treated as movement around that line rather than part of it. through the observed output series and calls the trend potential. The quiet assumption there is that the average of the past was itself sustainable. Another builds potential up from estimates of the workforce, the capital stock and how effectively they combine. Building it that way requires estimating each of the three, then estimating how they interact. Neither method is unreasonable and they do not agree.

Here is the demonstration, and it is worth slowing down for. Sankhya year 3 produced Rs 17,47,200 crore, and that number is not in dispute at any point. With potential growing at a steady 6.00 per cent from Rs 16,00,000 crore in year 1, potential reaches Rs 16,96,000 crore, then Rs 17,97,760 crore, and the year 3 gap is minus 2.81 per cent. Take potential growing at a steady 4.50 per cent instead, an assumption that is neither extreme nor silly. Rs 16,00,000 crore compounds to Rs 16,72,000 crore and then to Rs 17,47,240 crore. Actual output is Rs 17,47,200 crore. The shortfall is Rs 40 crore on a base of over seventeen lakh crore, or minus 0.0023 per cent, and it rounds to nothing at all.

Same year, same actual output that nobody disputes, and the gap is either minus 2.81 per cent or nil, depending purely on an assumption about capacity that was never observed and never can be. One and a half points of assumed potential growth is the entire difference between an economy carrying visible unused capacity and an economy sitting exactly where it should be. Nothing was produced differently in the two readings. Nobody counted anything differently. Only one number changed, the capacity growth rate somebody chose.

How can an output gap for a past year change years after the year ended?

Because the estimate of potential gets redone. A statistical system revises its output series as better records arrive. The whole series is also reset periodically against a fresh base year, in what is called a benchmark revisionA periodic overhaul in which a statistical system rebuilds a series on updated sources and a new base year, so levels and growth rates for years already published can move.. Anyone estimating potential rebuilds it on top of whatever the current output series says. Change the inputs and the estimated capacity line moves, and every gap computed against it moves with it, including gaps for years that ended a long time ago.

Sankhya year 2 shows how far that can go, using only the two potential paths already set out. On the 6.00 per cent path, year 2 actual output of Rs 16,80,000 crore sits against potential of Rs 16,96,000 crore, so the gap is minus 0.94 per cent and the economy looks to have unused capacity. Re-estimated at 4.50 per cent, year 2 potential becomes Rs 16,72,000 crore. Actual output has not moved by a single rupee. Output is still Rs 16,80,000 crore, and it now stands Rs 8,000 crore above potential, so the gap is plus 0.48 per cent and the economy looks to have been running slightly hot. The sign reversed on a year that was over, with nothing about that year re-counted.

There is a cruel pattern inside this. Confidence at the time comes from a settled view of capacity, and a settled view is exactly what the next rebuild unsettles. So the years for which a gap is quoted most confidently are often the years revised hardest later. A gap that felt obvious is not more durable than one that felt uncertain, and is often less.

What does an analyst actually do with a published output gap?

Treat it as a rough statement about how much room an economy has, and never as a precise quantity. The practical routine has three steps. First, find out who produced it. A gap is the property of a method and not of a country. Second, find out what potential path that method assumed. As the Sankhya comparison shows, the assumed growth of capacity carries more of the answer than anything counted. Third, find out what vintageWhich released version of a series a figure was computed from. The same year can carry several vintages, and two numbers built on different ones are not comparable. of the output series it was computed on, because a gap built on a superseded series is describing a year that has since been renumbered.

A gap quoted to two decimal places is claiming a precision that the concept does not have, and the sensible response is to read the first digit and treat the rest as decoration. Minus 2.81 and minus 2.89 are the same statement about Sankhya. So, honestly, are minus 2.81 and minus 1.9, if the difference between them is one analyst assuming 6.00 per cent capacity growth and another assuming 5.5. A reader can take two things from a gap, the sign and the rough size: is there room, and is it a little room or a lot.

The same discipline shows up when two published gaps for one year disagree. The first instinct is to decide one of them is wrong. The disagreement almost always lives in the two potential paths, so the better instinct is to line them up side by side. Seeing them together teaches something the two headline numbers were hiding. A lending team reading a note about capacity utilisationThe share of an installed production capacity that is actually being used, usually measured plant by plant through surveys. Utilisation is a direct observation of idle equipment rather than an estimate of an economys sustainable output. across an industry is on firmer ground than one reading an economy wide gap, because utilisation surveys ask factories what share of their installed capacity is running and get an answer about equipment that exists.