Sector rotation and the business cycle
The answer
Sector rotation is the observation that different parts of the share market lead at different points in the economic cycle. The pattern is real enough to be visible in long-run history and it is far too loose to time. The reason is simple: the phase of the cycle you are in is only identified with certainty long after the decisions that depended on it had to be made.
Why this costs you money
Two mistakes, and most readers make the second one without ever hearing of the first.
The first mistake is trading the clock. You read that the economy is in late expansion, so you sell technology and buy energy and materials. Six months later official data is revised and it turns out the expansion had already ended. Your positioning was for a phase that was no longer running. You did not make an analytical error. You used a map whose "you are here" marker is always placed retrospectively.
The second mistake is the expensive one, and it is passive. You never think about sectors at all, and you end up concentrated in 1 without noticing.
This happens naturally. You buy the companies you understand or the ones that have been doing well. Over 3 years that produces a portfolio of 8 holdings that looks diversified because it contains 8 names. Then you group them by sector and find 5 are financial companies and 2 are software companies. Your 8 decisions were 2 bets.
In India this is close to a default outcome, because the largest and most familiar companies are concentrated in a small number of sectors, and because the main index is itself heavily weighted toward financial services. You can build a portfolio that mirrors the index's concentration without ever intending to.
The action at the end of this article is the fix and it takes 5 minutes. Most people who do it find 1 sector holding more than 40% of their money.
How it works
The classic clock
The standard framework divides the economic cycle into 4 phases and assigns sector leadership to each. Sam Stovall set out a well-known version in the Standard & Poor's Guide to Sector Investing in 1996, and asset managers publish their own versions, including a widely circulated series from Fidelity.
| Phase | What the economy is doing | Sectors usually said to lead |
|---|---|---|
| Early expansion | Recovering from recession, rates low, credit reopening | Consumer discretionary, financials, real estate, industrials |
| Mid expansion | Growth steady, profits rising, capacity being added | Information technology, industrials, communication services |
| Late expansion | Growth slowing, inflation and rates rising, costs up | Energy, materials, consumer staples |
| Recession | Output contracting, unemployment rising, rates being cut | Consumer staples, health care, utilities |
The logic behind each cell is real and worth stating, because that is the part you can transfer.
Interest-rate sensitivity. Housing, cars and capital equipment are bought with borrowed money, so their demand responds first when rates fall. That is why consumer discretionary and real estate are placed early.
Operating leverage. Companies with high fixed costs show large profit increases when volumes recover, so their earnings rise fastest in the early and middle phases.
Input costs. Commodity producers benefit from the price increases that appear late in an expansion, which is why energy and materials sit there.
Demand that does not change. People buy food, soap and medicine in a recession. Companies selling those have stable revenue, so their shares fall less when profits elsewhere are collapsing. That is what "defensive" means, and it is the only sector claim in the table with a straightforward mechanism.
The problem in one sentence
The phase is a label applied afterwards.
In the United States, recessions are dated by the Business Cycle Dating Committee of the National Bureau of Economic Research. The committee waits for revised data before announcing a turning point, and the delay is often more than a year. This is not a criticism of the committee. Its purpose is to produce an accurate historical record, not a real-time signal.
But it means the clock's input is unavailable when you need it. If professional economists with full access to the data cannot confirm a turning point for 12 months, a reader deciding between financials and utilities on a Tuesday morning has no reliable way to know which phase they are in.
What is left after you remove the timing claim
Three things, and they are all usable.
1. Sectors do behave differently, and the differences are large. The dispersion between the best and worst performing sector in a year is routinely 30 to 60 percentage points in both markets. Whatever you think of the clock, sector allocation is a large part of your outcome.
2. Leadership persists for months once it establishes. This is not a business cycle claim. It is the momentum effect applied to groups instead of companies, and it has a much better evidence base than the clock does. You can observe current leadership directly by ranking sector indices on 6-month relative performance. That requires no view about the economy at all.
3. Defensive sectors really are defensive. Consumer staples, health care and utilities fall less in declines. The mechanism is stable demand and it does not depend on identifying a phase.
Notice what has happened. The forecasting part of sector rotation is weak. The measurement part is strong and free. That is the reverse of how the subject is usually taught.
The definitions move underneath you
One technical point that invalidates a lot of published sector history.
The Global Industry Classification Standard, used for American and global sector definitions, has been revised several times. Real estate was separated out of financials into its own sector in 2016. A communication services sector was created in 2018, which moved several very large companies out of information technology and out of consumer discretionary.
So a chart showing "technology sector returns since 1990" is showing a series whose contents changed. Any backtest of a rotation strategy across those boundaries is comparing categories that are not the same categories.
What it tells you, and what it does not
Sector rotation tells you where your money is concentrated by economic driver. Group your holdings by sector and you learn what single event would damage most of your portfolio at once. That is a risk measurement and it is reliable.
It tells you which groups are currently leading, if you measure rather than predict. Ranking sector indices by 6-month change against the broad index is a 5-minute task and it is a fact rather than a forecast.
It does not tell you which phase the economy is in. Nobody knows that in real time, including the people whose job is to decide it.
It does not tell you that a sector will lead because a phase says it should. The 2020 experience is the clearest counter-example available and it is in the cases below.
It does not transfer between countries. The clock was built on American sector definitions, an American cycle and an American central bank. India's sector composition is different, its cycle is driven by different things, and its policy calendar is different. Applying the American clock to Indian sector indices produces recommendations for sectors that either do not exist here at meaningful size or behave differently.
The decision rule
Use sector analysis for measurement first and for positioning second.
Step 1, always. Group your holdings by sector and compute each sector's share of your portfolio value. Any sector above 30% is a concentration, whether you chose it or not.
Step 2, if you want to position. Rank the sector indices by 6-month change against the broad index. Act on what is leading now, which is observable, not on what the phase says should lead, which is not.
Unless the leading sector is leading because of a one-off event rather than a trend — a single regulatory change, a single commodity spike, a single large constituent. Check what is inside the sector index before treating the ranking as information.
Never rebalance a portfolio on the basis of a phase call. If the phase cannot be confirmed for a year, a position taken on it cannot be evaluated for a year either.
Try this now
Five minutes, on your own holdings. Most people are surprised by step 4.
- Open your holdings list. Write down each holding and its current value, not the quantity and not the profit.
- Next to each, write its sector. Use the broad label your exchange or your broker uses: financials, information technology, energy, materials, industrials, consumer staples, consumer discretionary, health care, utilities, communication services, real estate. If you do not know a holding's sector, that is itself a finding, and you can look it up on the company's exchange page in under a minute.
- Add up the value in each sector. Divide each sector total by your total portfolio value including cash. You now have a percentage per sector.
- Sort the sectors by percentage, largest first. Look at the top 2 numbers and add them together.
- Now look at the table earlier in this article. Which phase does your sector mix correspond to? Write down the phase your portfolio is positioned for.
- Write 1 sentence saying whether you chose that positioning.
What you should see. For most readers the top 2 sectors add to somewhere between 50% and 75% of the portfolio. That is a concentration, and almost nobody arrived at it deliberately. Positions accumulate from what was familiar and what was performing, and both of those cluster.
Step 5 is the part people remember. Many portfolios turn out to be positioned for a phase the owner would not have chosen if asked directly. A portfolio that is 60% financials and industrials is positioned for early expansion. A portfolio that is 55% consumer staples and pharmaceuticals is positioned for a recession. Neither owner sat down and decided that.
The point is not that you should now rearrange it to match a phase. The point is that you had a macroeconomic position all along and you had not measured it. Now you have.
If your top sector is above 40%, do 1 more calculation. Work out what your portfolio would be worth if that sector fell 30%, which sectors regularly do. That number is the real question this exercise asks.
Three real cases
1. The 2020 recession and the announcement dates — the lag that makes the clock unusable The National Bureau of Economic Research determined that a peak in United States economic activity occurred in February 2020, and announced that determination on 8 June 2020. It later determined that the trough occurred in April 2020, and announced that on 19 July 2021.
Read the second one again. The end of the recession was officially identified roughly 15 months after it happened. During those 15 months American share prices rose substantially.
A person waiting for confirmation of the phase before positioning for early expansion would have positioned for it more than a year into the expansion. A person not waiting for confirmation was guessing. There is no third option, and this case is the reason this article treats the clock as description rather than method.
2. Calendar year 2020, United States sector returns — the phase said one thing and the market did another 2020 contained a severe, if short, recession. The clock says recession favours consumer staples, health care and utilities, and disfavours consumer discretionary and technology.
What happened was the reverse. Information technology and consumer discretionary were among the strongest sectors for the calendar year, while utilities and energy were among the weakest.
The explanation is not that the clock was applied badly. It is that the dominant driver in 2020 was not the ordinary business cycle. It was an enormous monetary and fiscal response, plus a demand shift toward companies that benefited from people staying at home. When a shock of that kind arrives, sector leadership is decided by the shock and not by the phase.
That is the general lesson and it applies beyond 2020. The clock assumes the cycle is the dominant driver. In many periods it is not.
3. Calendar year 2022, Indian sector indices — real rotation, wrong explanation In 2022 the 2 ends of the Indian sector table changed places. Nifty IT fell heavily while Nifty PSU Bank rose strongly.
This looks like textbook rotation and it was driven by 2 things that have nothing to do with the Indian business cycle. Nifty IT is dominated by companies earning revenue from clients in the United States and Europe, so it fell with the global technology spending cycle and with rising global interest rates. Nifty PSU Bank rose because the balance sheets of state-owned Indian banks improved after years of provisioning against bad loans, and because rising rates widen the margin between what a bank earns and what it pays.
The rotation was real, large and visible in relative strength months before the year ended. The explanation for it was not the Indian economic phase. This is the Indian version of the general point: measure the leadership, and be careful about the story you attach to it.
The question that resolves it
A novice asks: which phase of the cycle are we in?
An expert asks: would I be able to tell the difference between this phase and the next one, using only data available today?
That reframing kills most sector-rotation arguments in about 10 seconds. Late expansion and early slowdown look nearly identical in real time. So do late recession and early recovery. The data that separates them — revised output figures, revised employment figures, profit data for a quarter not yet reported — does not exist yet on the day you have to decide.
An expert therefore uses the part of the framework that does not need the phase. Ranking sector indices by measured relative strength requires no economic call at all, and the ranking is available today.
What would make this wrong
This article claims the sector-cycle link is real in history and weak as a timing tool. Both halves are open to challenge.
The historical claim would be wrong if sector returns turned out to be unrelated to the cycle once you correct for the number of combinations tested. There are 11 sectors and 4 phases, giving 44 cells, and any historical dataset will fill those cells with something. A finding that energy leads late expansion needs to survive being asked how many other sector-phase pairs were examined before that one was reported. I have not seen a study of the clock that applies a multiple-testing correction.
The weak-timing claim would be wrong if somebody built a real-time phase indicator that identified turning points promptly, published it in advance, and recorded its errors. Several institutions publish such indicators. Their real-time records, as opposed to their back-fitted records, are the thing to ask for, and they are rarely offered.
The strongest defence of the clock, stated fairly. Nobody serious claims it times turns. The claim is that it provides a prior — a default expectation that you adjust with other evidence. That is a reasonable and modest use, and this article does not object to it. What it objects to is the version sold to retail readers, which presents the clock as a positioning system with a current phase reading attached.
And the honest limit of the criticism. Defensive sectors falling less in declines is not a data-mined result. It follows from the stability of demand for food, medicine and electricity, and it shows up in essentially every downturn in both markets. That piece of the framework needs no phase identification and it survives every objection here.
In India
Indian sector rotation looks different from the American version for 4 structural reasons, and each one changes what you should do.
Financials dominate. Financial services carry a very large weight in the Nifty 50, far larger than any single sector's weight in the S&P 500. That has 2 consequences. Any statement about "the Indian market" is largely a statement about banks and lenders. And a portfolio that tracks the index is not sector-diversified in the way an American index portfolio is.
Several clock sectors barely exist here. The Indian market has no large, liquid utilities sector comparable to the American one, and real estate is a very small share of the index. Consumer staples exist mainly as fast-moving consumer goods companies. So an Indian investor asked to rotate into utilities and real estate for a particular phase has almost nothing to rotate into.
Sector indices are highly concentrated. Nifty IT is dominated by a handful of companies and the largest 3 account for a very large share of it. Buying "the IT sector" in India is closer to buying 3 companies than to buying a sector. This matters because the whole point of sector allocation is to take a group exposure without company risk, and in India that separation is much weaker.
The cycle itself is different, and less synchronised with the United States than commentary assumes. India's growth has been driven substantially by domestic consumption and investment rather than by exports of goods. India did not experience a recession in the standard sense during the global downturns of 2001 or 2008, though growth slowed. The one clear contraction in the modern record was the year affected by the pandemic, when annual output fell. The Reserve Bank of India's rate cycle has repeatedly diverged in timing from the United States Federal Reserve's.
Two Indian calendar effects are worth adding, because they create real annual patterns that have nothing to do with the business cycle. The financial year runs from April to March, so results, tax planning and fund flows follow a different annual rhythm. And the Union Budget, presented at the start of February, moves specific sectors on a specific date every year for policy reasons.
In the United States
The United States is where this framework was built, and it works there better than anywhere, which is not the same as working well.
The sector structure is clean and purchasable. Eleven GICS sectors, each with a large, liquid, low-cost exchange traded fund. An investor can express a sector view at a spread of a cent or two and take no company risk at all. That is the cleanest possible implementation of the idea and it does not exist in most other markets.
The data is timely and detailed. Monthly employment data, monthly and quarterly output data, weekly claims data, published survey indicators, and a policy rate path priced explicitly in futures markets. If real-time phase identification were possible anywhere, it would be possible here.
And it is still not reliable. Recession dating still takes months to years. Yield curve inversion, the most watched single recession indicator, has led recessions by variable and sometimes very long periods, and produced at least one widely discussed signal that was not followed by a recession on the expected timetable.
The sector definitions changed twice recently. Real estate was separated from financials in 2016 and communication services was created in 2018. Any American sector rotation backtest spanning those years is not comparing like with like.
Sector concentration inside the index has grown. The largest American companies have become a very large share of the S&P 500, and they are concentrated in a small number of sectors. That means American index investors now face a version of the concentration problem that Indian investors have always had, though for different reasons.
Where they differ, and what that tells you
Three differences, and each one changes the method rather than decorating it.
The clock's sector list does not map onto India. Four of the phases in the American clock point to sectors that are either absent, tiny or extremely concentrated in India. That is not a translation problem. It means the clock cannot be run here in the form it is taught, and any Indian course teaching it unchanged is teaching an American framework with Indian labels on it.
India's cycle is less synchronised with the United States than commentary suggests, but India's asset prices are more synchronised than its economy. This is the important and counterintuitive part. Indian growth is driven substantially by domestic demand, so the real economy can diverge from the American one. Indian share prices are set at the margin partly by foreign portfolio flows, which respond to American interest rates. So you can have an Indian economic cycle that is out of step and an Indian equity market that moves with global risk appetite anyway. Positioning for the Indian phase can therefore be right about the economy and wrong about the market.
The measurement approach transfers and the forecasting approach does not. Ranking sector indices by 6-month relative performance works identically in both markets, because it requires no view about the economy. That is the part an Indian reader should take. The phase-based positioning is the part to leave, and the reason is not that Indians should be more cautious. It is that the framework's inputs are less available and its sector categories do not exist here.
Carry this
- The clock is a description of history. Its input, the current phase, is not available in real time.
- Rank sector indices by measured 6-month relative performance instead. That is a fact, not a forecast.
- Group your own holdings by sector before anything else. Most portfolios have an unintended 50% to 75% in 2 sectors.
- Defensive sectors falling less in declines is the one part of the framework with a mechanism that never needs a phase call.
- India has no meaningful utilities or real estate sector to rotate into, and its sector indices are 3 or 4 companies wearing a group's name.