The economic cycle: cyclical, defensive and growth companies
Now let's use everything we learned on three famous companies that behave very differently when the economy changes. Same tools, three completely different stories.
Caterpillar (CAT)
Construction and mining machines, engines and turbines. Sales follow the economy up and down.
Procter & Gamble (PG)
Tide, Pampers, Gillette, Oral-B. Things people buy every week, whatever the economy.
Microsoft (MSFT)
Cloud computing (Azure), Office, Windows, LinkedIn, Xbox. Mostly subscriptions.
1. Revenue: the ride matters, not just the destination
The chart sets each company's 2019 revenue to 100, so you can compare their paths regardless of size.
| Revenue ($bn) | 2019 | 2020 | 2021 | 2022 | 2023 | 2024 | Yearly growth (CAGR) |
|---|---|---|---|---|---|---|---|
| Caterpillar | 53.8 | 41.7 | 51.0 | 59.4 | 67.1 | 64.8 | 3.8% |
| Procter & Gamble | 67.7 | 71.0 | 76.1 | 80.2 | 82.0 | 84.0 | 4.4% |
| Microsoft | 125.8 | 143.0 | 168.1 | 198.3 | 211.9 | 245.1 | 14.3% |
Surprise: Caterpillar and P&G grew at almost the same average speed (about 4% a year). But Caterpillar's sales fell 22% in 2020 while P&G's rose 5%. An average hides the ride — which is why our analysis pages always show the worst year and the volatility of growth next to the average.
The 2009 financial crisis tells the same story: Caterpillar's sales dropped by about 37% (from $51.3bn to $32.4bn), P&G's by only about 3%.
Revenue change in the 2020 shock
2. Profit: operating leverage in action
| Net income ($bn) | 2019 | 2020 | 2021 | 2022 | 2023 | 2024 |
|---|---|---|---|---|---|---|
| Caterpillar | 6.1 | 3.0 | 6.5 | 6.7 | 10.3 | 10.8 |
| Procter & Gamble | 3.9* | 13.0 | 14.3 | 14.7 | 14.7 | 14.9 |
| Microsoft | 39.2 | 44.3 | 61.3 | 72.7 | 72.4 | 88.1 |
* P&G's 2019 profit was cut by an $8.3bn non-cash write-down of the Gillette business — a one-off item (see the income statement lesson: always check for one-offs).
Net margin (net income ÷ revenue)
Caterpillar: revenue fell 22% in 2020 but profit fell about 51%. Factories, engineers and dealers cost money even when fewer machines are sold — remember the bakery quiz on operating leverage? It works in both directions: when demand came back, profit grew much faster than sales (net margin from ~7% to ~17%).
P&G: net margin stays in a narrow band around 17–19%. Strong brands let it raise prices when its own costs rise.
Microsoft: margins above 30%. Selling one more software subscription costs almost nothing, so most extra revenue becomes profit. Building data centres for the cloud, however, now requires very large capital spending — watch capex in its cash flow statement.
3. Side by side
| Caterpillar | Procter & Gamble | Microsoft | |
|---|---|---|---|
| Business model | Sells expensive machines customers can postpone buying; services and parts are steadier | Sells cheap everyday necessities, bought again and again | Sells software and cloud services, mostly as recurring subscriptions |
| Revenue in bad times | Falls sharply (−22% in 2020, −37% in 2009) | Barely moves | Kept growing in 2020; dipped slightly in 2009 |
| Net margin (2024) | ≈ 17% (≈ 7% in 2020) | ≈ 18% | ≈ 36% |
| Debt | Looks high, but much of it funds loans to customers buying machines (Cat Financial) — read the notes before comparing | Moderate; steady cash flow makes debt easier to carry | Low compared with its cash flow |
| Dividends | Raised every year for more than 25 years | Raised every year for more than 60 years | Paid since 2003, raised regularly |
| What drives it | Construction, mining and energy investment; commodity prices; interest rates | Population, prices, brand strength; costs of raw materials and currencies | Cloud and AI demand; corporate IT budgets; competition |
4. How the type of company changes the way we read valuation
Cyclicals and the P/E trap. At the top of the cycle profits are at their highest, so the P/E looks low. At the bottom, profits shrink and the P/E looks high. For a cyclical company a single year's P/E can therefore point in the opposite direction from what beginners expect. Analysts often average earnings over a full cycle (e.g. 7–10 years) instead.
Defensives and stable multiples. Because profits are steady, P/E and EV/EBITDA tend to move in a narrower range, and investors often accept higher multiples in exchange for that stability.
Growth companies and the DCF. Most of the value lies in profits many years away, so the answer depends heavily on the growth rate and discount rate you choose (see the DCF lesson, step 9).
For a DCF on any of them, using a 10-year history captures a fuller cycle than the last 2–3 years. That's why our analysis pages use the 10-year average growth as the default — and let you change it.
5. What to take away
- Always look at 10 years, not one. A single year can be a peak or a trough.
- Ask "what happens to this company in a recession?" The answer shapes everything else.
- Similar averages can hide very different risks — check the worst year and the volatility.
- There is no "best" type: cyclicals, defensives and growth companies each fit different goals and temperaments.
Show the answer
No. Steel is very cyclical. A record year often means profits near the top of the cycle; if profits fall by 70% the next year, the "real" P/E at today's price would be about 17. Look at earnings over the whole cycle.