Secular Growth: Demand That Does Not Depend on the Cycle

Secular growth is demand that rises for reasons that do not reverse with the economy: more households, a changed habit, a replacement cycle that has shortened. Cyclical growth rises and falls with activity and comes back. The two look identical in any single year. The distinction is made from a record rather than from a number.

Almost everything difficult about this idea comes from one awkward fact. The distinction is about causes, and causes are invisible. A number is what gets handed over, and a number carries no label saying which of the two produced it. Sorting growth into the right box on sight is therefore not the work. The work is knowing what kind of evidence could ever do the sorting, and saying plainly when that evidence is absent.

What is secular growth, and what does the word actually claim?

Secular growth is demand rising because something in the world changed and stayed changed. More separate households exist than existed ten years ago, so more kitchens need painting. A habit shifted, so a job people once did every ten years they now do every seven. A product moved from something a few people bought to something most people buy. In each of those, the reason the volumes are higher this year is still sitting there next year, whatever the economy is doing.

A secular claim must name the structural change, or it is a description of some growth rather than a claim about the kind of growth it is. That is a stricter test than it sounds, and most sentences written about structural demand fail it. Saying volumes have grown strongly for three years describes a record. Saying demand is structural asserts a cause. If the sentence does not contain the cause, it has borrowed the authority of the second while doing only the work of the first.

Take the everyday version. A tea stall outside a college gate sells more cups this year than last. Two quite different worlds produce that. In the first, the college took a second intake and there are simply more students on the road every morning, permanently. In the second, a building nearby is under construction and the workers come by for six months. The stall sees the same higher number in both. Only one of them is still there next year, and no amount of staring at the day's takings tells the owner which one he is in.

The invented field this material uses is decorative and industrial coatings, and its secular part is described as household formationThe rate at which new separate households come into existence, whether by people moving out, marrying, or migrating to a city. Household formation is a demographic measure rather than a financial one. together with a repainting cycle that has been getting shorter. The shape of the claim matters more here than its truth: it names a change, and the change is one that does not undo itself when activity slows.

Cyclical Growth: what is it, and why is it not the inferior kind?

Cyclical growth is demand rising because activity is high right now. New building starts are running hot, so more surfaces need coating. Factories are adding lines, so more industrial coating gets bought. Then the same forces run the other way, orders thin out, and the volumes come back down. The rise was completely real while it lasted. The rise lacked any reason to persist once the conditions that produced it stopped holding.

Cyclical growth is real growth and not the lesser sort. The only thing it lacks is persistence. This is worth insisting on, because the words carry a snobbery they have not earned. Money made in a cyclical upswing spends exactly like any other money. Volumes sold in a strong year were genuinely sold. A field that swings is not a worse field, it is a differently shaped one, and describing it accurately is the whole job. The distinction exists to indicate what to expect next, not to rank the two.

Size and quality do not separate the two. Behaviour when the weather turns does. Secular demand carries on because the reason for it is still there. The activity was the reason, so cyclical demand goes where the activity goes. In the invented field here the cyclical drivers are new construction and industrial capital spendingMoney a business commits to building or buying long lived assets such as plant, machinery and buildings, as against the money it spends running the operation day to day., both of which are famously willing to stop for a year and then start again.

Why can a single year never settle which of the two it is?

Because in any single year both kinds of demand do exactly the same thing. Each produces a higher number. There is no second column beside the figure that says whether the extra volume arrived for a reason that persists. The cause lives outside the number, and the number is all that was given.

A reader shown one year of growth has no information at all about which kind it is, and this does not improve by looking at that year more carefully. That last clause is the part people resist. There is a strong instinct that a bigger number, or a cleaner number, or a number broken into more pieces, must eventually reveal its own cause. It cannot. Decomposing 6.8 per cent into volume and price, or into regions, or into channels, shows where the growth sat. A decomposition says nothing whatever about whether the conditions that produced the growth will still be true in eighteen months.

Given one year, the honest answer to what kind of growth is this is that the question is not yet answerable. Not that it is probably secular because the field seems to be changing, and not that it is probably cyclical because growth like that never lasts. Both of those are guesses dressed as readings. The evidence that separates the two is not in a single year, and no technique applied to a single year puts it there.

How much record is actually needed before the two separate?

Enough of a record to contain at least one bad year. One bad year is the whole answer, and it is a very different answer from a number of years. Behaviour under stress tells the two apart: what happened to the volumes when activity fell away. The reason for secular demand is still standing, so secular demand holds up better than the activity around it. Cyclical demand goes down with the tide. If the record contains no year where the tide went out, the distinguishing behaviour has never been observed.

The requirement is therefore about what the record contains rather than how many years it has, and ten years without a downturn separates nothing at all. A decade of uninterrupted growth feels like overwhelming evidence and is, on this specific question, evidence of nothing. Ten good years are ten observations of a field during good conditions. The one observation that was needed is missing, and the length of the run gives a false sense that the matter has been settled by weight of data.

The household version is exact. Whether a person is a careful spender cannot be established by watching them in a month where money was plentiful. Everyone looks careful when nothing is tight. The month that settles it is the one where the salary came late. Four years of a record containing one genuinely bad year is worth more, for this question, than fifteen years of steady prosperity.

What makes a secular claim testable rather than a story?

A claim that can be acted on has three parts, and dropping any one of them turns it back into a story. The claim names the structural change. The claim says what would be seen if that change were real. And it says what would be seen if it were not. The third part is the one almost always missing, and it is the one doing the work. A claim with no way of being wrong cannot be checked by anyone, including the person who made it.

A secular claim that cannot be wrong is a story rather than an analysis. The same checkability requirement governs every view in this material, applied here to a claim about demand. Nothing new is being introduced here. A view has to be stated so that reality can contradict it, or it is not a view. Words like structural and long term point away from any particular year in which they might be tested, so secular claims are unusually easy to phrase in a way that quietly forbids contradiction.

Placed side by side, the two are easy to tell apart. Demand in this field is structurally rising is unfalsifiable as written: any bad year is absorbed as noise, any good year is confirmation. The repainting cycle has shortened by about three years and stayed shortened, so volumes should hold within a few points of trend even in a year when construction falls, and if volumes fall with construction the claim is wrong is a claim. The longer version names the change, says what holding up looks like, and supplies the observation that would kill it.

What does the invented five year record establish, and what does it not?

Here is the record this material uses for the field. Volume growth of 2.1 per cent, then 6.8 per cent, then a fall of 1.4 per cent, then 7.2 per cent, then 4.5 per cent. The five add to 19.2, so the arithmetic meanThe plain average: add the values and divide by how many there are. The mean weights every observation equally and is not the same as a compound rate over the same span. is 3.84 per cent. The deviations from that mean run from plus 3.36 to minus 5.24 percentage pointsThe unit for the gap between two percentages. Going from 4 per cent to 6 per cent is a rise of two percentage points, and also a rise of fifty per cent. The unit has to be said aloud., and by construction they add back to zero.

The record establishes that a contraction occurred inside a field that grew over the period. A cycle sitting on top of a trend looks exactly like that. Volumes ended the five years higher than they started and they did not get there in a straight line. Something pulled them down in year three and something pulled them back in year four. The fall and the recovery are the signature shape, and they are genuine evidence: a field with no cyclical component would not produce a minus 1.4 in the middle of it.

The record does not establish the size of either part. Five points cannot separate a trend from a cycle with any precision, and the mean of five noisy numbers is not the secular rate of anything. The mean is the average of what happened to occur in those particular five years, and most of that is an accident of where the window was cut. Worth noting too that the mean is not even the right kind of average for a growth rate: compounding the five years and taking the fifth root gives 3.79 per cent, a shade below the 3.84 per cent the simple mean reports.

Now the demonstration that matters. The two strongest years of the record are 6.8 and 7.2 per cent. A reader shown only those would describe a field growing at 7.00 per cent. The three weakest years are 2.1, minus 1.4 and 4.5 per cent, adding to 5.2. A reader shown only those would describe a field growing at 1.73 per cent. The two readings are 5.27 percentage points apart and they come from the same five numbers. Nobody has to lie or fabricate anything to get from one to the other. Choosing which years to show is enough.

Sitting behind the record is the structural claim itself, and it has to be stated as a claim. The invented driver is a repainting cycle that has been shortening. A rising share of volume going to repainting rather than to first coats on new surfaces would show it, and would have to hold up in a year when construction falls. The case record here does not contain that evidence. So the claim stays open: not endorsed, not dismissed, and named as the thing to go and look for next.

How does a secular trend actually end?

Three ways, and most treatments skip all of them. The trend saturates: nearly everyone who was going to take up the habit has taken it up, and the pool of new adopters thins out. The structural change completes: a repainting cycle that shortened from ten years to seven does not go on shortening forever, and once it stops moving it stops adding growth. Or something substitutes for it: a different material, a different method, a coating that lasts twice as long and halves the repeat purchase.

A secular trend ending looks exactly like a cyclical downturn while it is happening, so the same problem recurs at the other end of the trend and is considerably harder there. Harder for a specific reason. At the start of a trend the analyst is only being asked to withhold judgement. At the end a judgement has already been formed, written down, defended, and possibly built into a whole forecast. Now the same ambiguous falling numbers arrive, and the reading that says this is just the cycle costs nothing while the reading that says the trend is over costs the position already taken.

The two cannot be told apart from the falling numbers, and the attempt is not worth continuing. The mechanism separates them. The structural change that was named has to be revisited and tested for whether it is still running. Has the repainting cycle stopped shortening? Has the pool of unpainted new households stopped growing? Both are questions about the world rather than about the series, and both are answerable. The series is not going to answer them at either end.

What does confusing the two actually cost?

Both directions are expensive, and they are expensive in opposite ways. Cyclical growth treated as secular takes the top of a swing and projects it forward as though it were the ordinary state of the world. The forecast is then wrong by the entire cyclical component. In the record above that component is worth more than five percentage points a year. Secular growth treated as cyclical leaves the analyst waiting for a reversal that is not coming, hedging against a downturn that the structural change has removed, and wrong for a very long time in an expensive and quiet way.

Both errors are costly, and the first is far more common. A good year is simply more pleasant to extrapolate than a bad one. That is not a technical failure. Nobody misreads a cycle for a trend because the arithmetic defeated them. The misreading happens because the strong-year reading produces a better looking number, and because extrapolationContinuing a pattern beyond the data actually in hand, by assuming that whatever produced it goes on producing it. The assumption is the whole of the risk. of a good run is what an audience wants to hear. The second error, waiting for a downturn that never arrives, is unpopular in the other direction and produces no headlines at all. Silence is part of why it survives so long uncorrected.

The second half of year four in the record is a small live example of the trap. After a fall of 1.4 per cent, the next year prints 7.2 per cent, the highest in the record. Some of that is a base effectA growth figure that looks large mainly because the year it is measured against was unusually small. Nothing about the level has to be remarkable for the percentage to be.: the year it is measured against was a shrunken one. Reading 7.2 per cent as evidence of a structural surge mistakes a rebound for a change in the world.

How does an analyst put this to work on a real morning?

Meghna Iyer covers coatings, and Sarvani Coatings Limited is on her list. When she opens the field's volume record she is not asking what the growth rate is. She is asking a narrower and much more useful pair of questions: does this record contain a bad year, and if it does, what did volumes do in it. Everything else she can compute later. If the answer to the first is no, she writes that down as a limitation of the evidence rather than working around it.

Her discipline is that the structural claim gets written as a sentence with a refutation attached before any number is carried forward, and no forecast leaves her desk resting on a record with no downturn in it. She does one more thing that is easy to skip. She separates the question about the field from the question about the company. Sarvani Coatings grew volume 6.0 per cent in the latest year against the field at 4.5 per cent. The gap is a company question about market shareThe slice of a field's total sales that one seller accounts for. How it is measured, and how small a real gain usually is, is taken up separately. and competitive position, taken up elsewhere. Whether the field's 4.5 per cent was trend or swing is the question in this guide, and mixing the two produces a sentence that answers neither.

A household does this correctly without being taught. Somebody deciding whether to expand a shop asks where the extra footfall came from. A new colony built next door is permanent. A nearby road shut for eight months, routing strangers past the door, is not. Nobody in that situation thinks the till receipts alone will settle it. The shopkeeper goes and finds out what changed. Finding out what changed is the entire method, and it does not become less valid when the field is larger and the record is in percentages.