The Business Cycle: The Four Phases and What Drives Each

A business cycle is the pattern output traces as it runs above and below the level an economy can sustainably produce. Four phases get named: expansion, peak, contraction and trough, with the climb out of a trough usually called a recovery. The pattern is not a schedule. No phase lasts a set number of years, no two are equal in depth, and naming the phase an economy is standing in is genuinely hard.

A business cycle rests on two numbers a reader already has. One is actual output, a figure somebody counted and published. The other is potential output. Potential output names a version of the same year carrying no idle capacity and no overwork anywhere in it, so nobody counted it and nobody could. Subtracting the second from the first and dividing by the second gives the output gap. A business cycle is what that gap shows once it is plotted across a run of years instead of read one year at a time. The phases are regions of that plot, and the arguments about phases are almost always arguments about the second number.

What is a business cycle, and what does the word cycle wrongly promise?

A business cycle is the swing of actual output in real termsProduction for the whole economy measured in constant prices, so a rise counts extra goods and services rather than dearer ones. around potential outputThe production level an economy can keep up using the labour, equipment and organisation already in place, without overworking any of it. Potential output is worked out from a chosen method, so two methods can disagree., repeated over years. Output rises faster than capacity for a while, then stops, then falls, then stops falling, then rises again. Each stretch has a name, and the names are the four phases. The distance between the two lines at any point is the output gapActual output less potential output, divided by potential output and written as a percentage. Positive means the economy produced more than the sustainable level, negative means less., and reading a cycle means watching that distance open, close and change sign.

Most readers arrive holding the wrong one of the two, so start with the difference between a cycle and a season. A caterer knows the wedding season arrives every year on dates the calendar hands out in advance. Bookings climb, staff are hired for the stretch, and the quiet months are budgeted for before they come. The caterer's year is a season, and a season is a schedule. A scooter dealer on the same street has no such calendar. Sales run strong for a few years, thin out over one bad stretch, and pick up again on no announced timetable, and the dealer discovers each turn by watching the showroom rather than the calendar. A business cycle behaves like the dealer's experience and not like the caterer's, and that is worth fixing in mind before any of the phase names arrive.

Now the harder half. The word cycle quietly promises something it cannot deliver. A cycle in ordinary speech means a wheel: fixed period, fixed amplitude, back to where it started. Nothing in the mechanism of an economy produces any of the three. The Republic of Sankhya, an invented economy, spent two years climbing above its potential and reached a gap of plus 4.37 per cent, then spent the following two years falling to a gap of minus 6.86 per cent. Same two years each way, and the fall is 2.57 times the size of the rise. Nothing makes the phases equal in length or equal in depth, so the word cycle names a repetition and promises a regularity that is not there.

Why does the Sankhya potential line grow at 4.00 per cent rather than 6.00?

Because a reader who has met Sankhya before has met a different capacity line, and the difference deserves a straight answer rather than a quiet substitution. The earlier treatment of Sankhya set potential growing at a steady 6.00 per cent a year. Consider what that does to an account of phases. Sankhya output grows 6.00 per cent in year 4, 6.50 in year 5, then falls, then rises 5.00 per cent in year 8. Against a capacity line climbing 6.00 per cent every single year, actual output never once gets above the line. The shortfall widens or narrows but never changes sign. An economy that can never rise above its own potential can never reach a peak, and an illustration that cannot show a peak cannot show four phases at all.

The honest fix was already sitting in the earlier work. Where potential output and the output gap are taught, the same year 3 was run a second time on a capacity line growing 4.50 per cent instead of 6.00. On that path, year 3 capacity works out at Rs 17,47,240 crore while the counted output for the year is Rs 17,47,200 crore. The two stand Rs 40 crore apart, and against a level above seventeen lakh crore that comes to 0.0023 per cent, so the gap is essentially nil and the economy is sitting level with what it can sustain. The earlier treatment made three claims, and all three are used here: potential is estimated rather than counted, two reasonable methods can differ by a wide margin, and an estimate for a year already finished can be rebuilt later.

So the capacity line here is not a convenience introduced to make the picture come out right; it is the earlier lesson being applied to a year that was always in dispute. The re-estimated year 3 level of Rs 17,47,240 crore is grown at 4.00 per cent a year through to year 8. The 4.00 per cent is itself an assumption, stated openly and held fixed. Holding it fixed keeps everything that moves in the picture coming from actual output. The rate is not measured and not observed, and a different assumption would move every gap figure here without moving a single output figure.

The Stages of a Business Cycle, and How a Year Is Placed in One: what happens in each stage?

Four stages, and each one is a statement about two things at once: what output is doing, and where output stands relative to capacity. An expansion is a stretch in which output is rising and the gap is widening in the positive direction, whether the gap has crossed above nil yet or not. A peak is a year the gap rose into and stops rising after. The year output stops rising is a different year. A contraction is a stretch in which output falls. A trough is the last year of that fall, after which output turns up again. The climb away from a trough carries its own name, a recovery, and it is the early part of the next expansion rather than a fifth stage.

Most readers slip on the peak definition, so watch it closely. A peak is not the year output was highest, and it is not the year the gap was largest for its own sake. A peak is the year the gap was still climbing into and stops climbing at, the last moment before the economy starts losing ground against its own capacity. Sankhya output in year 5 is Rs 19,72,414 crore, and output in year 8 is Rs 19,99,066 crore, which is higher. The gap at year 5 is plus 4.37 per cent and every gap after it is smaller, so year 5 is still the peak of this cycle. The peak is a statement about position against capacity, not a high score in rupees.

Take them in order on the Sankhya path and the shape becomes concrete. Year 3 sits level with capacity, where output of Rs 17,47,200 crore stands beside a capacity estimate of Rs 17,47,240 crore. Year 4 grows 6.00 per cent to Rs 18,52,032 crore while capacity grows 4.00 per cent to Rs 18,17,130 crore, so output pulls ahead by Rs 34,902 crore and the gap opens to plus 1.92 per cent. Year 5 grows 6.50 per cent and pulls further ahead to a gap of plus 4.37 per cent, the furthest it gets. Year 6 output falls 1.00 per cent while capacity keeps climbing, and the gap flips to minus 0.65 per cent. Year 7 falls another 2.50 per cent and the gap reaches minus 6.86 per cent. Year 8 output rises 5.00 per cent, so the fall has ended and the trough was year 7.

What drives each stage, and why is it one mechanism rather than four?

Begin on a street corner instead of inside an economy. A workshop opens and takes on eight people. The eight new workers now buy lunch every working day, so the tea stall on the corner sells more, and after a few months the stall keeps a helper on. The helper's household starts buying a second gas cylinder and gets the scooter serviced properly. The mechanic has a fuller week. Every one of those people is somebody else's customer, and every rupee of income is somebody else's revenue. The engine of an expansion is that spending creates income and income creates spending, so the movement supplies its own fuel and does not need a fresh push each year. Written for a whole economy, the sum of what everybody is trying to buy is called aggregate demandThe total spending on goods and services across an economy in a period, adding together households, businesses, government and net purchases from abroad. Aggregate demand is a sum of spending, not a measure of capacity., and an expansion is that sum running ahead of what capacity was quietly growing at.

So why does it ever stop? Because something binds, and there are three usual candidates. The first is physical. Plants, sites and staff run close to full, so an extra order can only be filled by working the same resources harder, and the strain shows up as rising capacity utilisationThe share of a plant, a fleet or an installed base actually in use, usually reported as a percentage of what could be run. A reading near the top says extra orders have to be met by straining rather than by switching idle capacity on.. The second is cost: when everybody is hiring and buying inputs at once, wages and input prices climb, and the same order is worth less to fill. The third is appetite for borrowing on either side. An expansion is usually financed, and both the willingness to lend and the willingness to take on more debt can turn well before anything physical binds. A peak is not a ceiling being hit; it is the moment the loop stops adding to itself, and any one of those three is enough to do it.

Then run the identical loop backwards. Orders thin out, so the workshop cuts the overtime it was paying. The eight households spend a little less, so the tea stall lets the helper go, so the mechanic loses a customer, so the mechanic's household defers a purchase. Nothing new has been introduced. The same chain that carried income into spending is carrying the absence of income into the absence of spending. A contraction is not a separate mechanism from an expansion. A contraction is the same mechanism run in the other direction. One explanation covers both, and a cycle is one thing rather than four.

A trough arrives the way a peak does, when the feedback stops feeding itself. Spending on food, rent, power and medicine has a floor and does not fall to nothing. Machines and vehicles wear out on their own schedule and eventually have to be replaced whatever the mood is. Shops that stopped ordering have been selling from the shelf, and once the shelf is bare somebody has to order again just to stand still. None of that is a rescue and none of it is timed. A trough is simply the point where the fall exhausts the material it was feeding on. Output turns up from there with the shortfall against capacity still wide open.

Which reading names a phase, the direction of output or the size of the gap?

Both, in a fixed order, and the order is what stops the whole exercise going wrong. Ask the direction question first: did output rise or fall this year? An economy producing less than it did last year is contracting even if it is still comfortably above its own capacity, so a fall is a contraction whatever the gap says. Only once output is rising does the gap get consulted, and then it is consulted twice: where does it sit, above nil or below, and is it still rising? A gap that is still rising means the economy is gaining ground on capacity, and gaining ground is an expansion. A gap that rose into the year and stops rising there marks the peak. A gap sitting below nil while rising is the early stretch of the same expansion, the part called a recovery.

A phase is decided by the direction of output together with where the gap sits, and never by the size of the gap on its own. Sankhya year 4 is the case that proves it. The gap is plus 1.92 per cent. A reader who has learned that a positive gap means an economy is running hot might reach for the word peak. Year 4 is an expansion, and it is one for a reason that has nothing to do with 1.92 being small: the gap goes on rising, to plus 4.37 per cent in year 5. Year 5 is the peak because the gap stops rising there, not because 4.37 is a large number.

Run the same test on the negative side and it holds. Year 6 has a gap of minus 0.65 per cent and year 8 has a gap of minus 5.96 per cent, so on size alone year 8 looks nine times worse. In year 6 output fell 1.00 per cent and in year 8 it rose 5.00 per cent, so year 6 is a contraction and year 8 is a recovery. Size ranked them in one order and the phase test put them in the opposite one. Size shows how far from capacity the economy is. The distance is worth knowing, and it answers a different question from which phase a year sits in.

One combination turns up constantly in real series, and the panel below produces it readily. Suppose output rises, but by less than capacity grew. Sankhya capacity climbs 4.00 per cent every year, so a year of 2.00 per cent growth leaves output higher and the gap lower at the same time. Output rose, so such a year passes the direction test. The gap was already falling into the year rather than rising, so the year fails the second test. The economy is growing and losing ground in the same twelve months, and none of the four names fits it cleanly. The honest description says exactly that and avoids the word peak. Peak needs a gap that was still climbing into the year it names.

The whole Sankhya path, with every year placed by the three tests

Here is the run in one place. Actual output starts from Rs 17,47,200 crore in year 3 and moves by the growth rate shown. Potential starts from the re-estimated Rs 17,47,240 crore and grows 4.00 per cent every year regardless of what output does. The gap is actual less potential, divided by potential, in per cent. The phase in the last column is the output of the three tests and not a label applied first and justified afterwards.

Output and potential are held in whole rupees in the arithmetic and shown rounded to the nearest crore, and gaps and growth rates are rounded to two decimals. The 4.00 per cent growth of potential is an assumption rather than an observation, so every gap in the table is read against a capacity line nobody measured.

The gap swings 11.23 points across this run, from plus 4.37 per cent at the peak to minus 6.86 per cent at the trough, and that swing is the only reason all four phases can be shown on one path. Two details in the table repay a second look. Year 8 produces Rs 19,99,066 crore against year 5's Rs 19,72,414 crore, so year 5 has the highest gap but not the highest output. And year 8, the recovery, sits 5.96 per cent below capacity. Year 6, in the middle of the contraction, sat at minus 0.65 per cent, a better position. Direction and position genuinely are two different readings, and the table shows them disagreeing in both directions.

Why is placing a year so hard from inside it?

Three reasons, and they stack. The first is that the output figure itself is late and provisional. A first estimate is built from partial returns and is replaced by fuller ones, so a year can be published as slightly up and later restated as slightly down. The restatement is a revisionA restatement of a figure that has already been published, made when better source data arrives or when the method is rebuilt. A revision changes the number for a period that is already over., and it is routine rather than a sign that anybody got anything wrong.

The second has already had a section of its own above. Potential is an estimate, so the position half of every phase call rests on somebody's method. Changing the assumed growth of capacity from 4.00 per cent to something else moves every gap in the table while every output figure stays exactly where it is. The third is the sharpest: the peak test asks whether the gap is still rising, and that question cannot be answered until the following year exists. Sankhya year 5 is only knowable as the peak once year 6 has been counted. A turn is visible from a year later and invisible from inside, and the limit is the shape of the test rather than a failure of anybody's judgement.

The three difficulties together give a conclusion worth stating plainly. Suppose the first published estimate for Sankhya year 6 had shown output up 0.30 per cent, and a later revision restated it as down 1.00 per cent. On the first estimate, year 6 passes the direction test as a rising year, the gap has stopped rising, and the careful reader names it the peak. On the revised estimate the same year fails the direction test and is a contraction. Nothing about production in year 6 changed. Two careful readers can place the same year in different phases and both be reasoning correctly. They are reasoning on different vintages of the same figure and on different estimates of the same capacity. Any account claiming that a phase can be named with certainty from inside it is selling a confidence the method does not have.

What does a lender actually do with a phase reading?

Not what most readers expect, and certainly not timing. A lender assessing a borrower cannot act on a phase call that will only be confirmed a year later, and would be taking on the forecasting risk of the whole economy if it tried. A phase reading is genuinely useful for a question about the borrower rather than about the economy: how quickly does a turn out there arrive here?

Two things decide that, and both are checkable from documents the lender already has. The first is who the borrower's customers are, and specifically how much of the order bookThe work a business has already been contracted to deliver but has not yet delivered and billed. An order book is a record of committed future revenue, and it says nothing about orders that have not been signed. comes from customers whose own spending moves with the cycle. A workshop supplying construction sites feels a turn within a quarter. Hospitals do not defer, so a firm supplying hospital consumables feels a turn much later and much less. The second is how long the contracts run. A borrower with three year contracts has three years of revenue a downturn cannot reach directly. A borrower selling one job at a time is exposed to the next quarter.

So the practical output is a lag estimate, not a timing call: this borrower feels a turn in about one quarter, that one in about five, and the covenant tests fall due in between. The Sankhya path shows why the distinction matters. Year 8 grows 5.00 per cent, and the growth reads as good news. Output is still Rs 1,26,719 crore short of capacity, and the shortfall is where the borrower's slack orders are sitting. A lender reading only the growth figure would mark the file as recovered while the shortfall in the borrower's own order book has barely started to close.