Case Study 1 — One Product, Two Interventions: The Reformulation That Worked and the Campaign That Didn't

⚠️ A composite constructed from documented policy episodes across several countries. ⚠️ The product is illustrative; the two intervention types, their outcomes and the arguments made against them are drawn from the real record described in §36.7 and §36.13b.**

Setup

⚠️ A single product category: a sweetened breakfast cereal aimed at children.

⚠️ Two things were tried, in the same country, roughly a decade apart.

⚠️ One of them changed what was in the box. ⚠️ The other did not.

Intervention A — the education campaign

⚠️ A publicly funded campaign encouraging parents to check sugar content on cereal labels and choose lower-sugar options.

⚠️ What happened:

⚠️ Awareness rose measurably. ⚠️ Parents surveyed afterwards were more likely to say sugar in cereal was a concern and more likely to report checking labels.

⚠️ Reported behaviour changed modestly. ⚠️ Actual purchasing changed very little. ⚠️ The composition of the product did not change at all.

⚠️ And the distributional finding, which is the one that matters:

⚠️ Uptake was highest among parents who were already buying lower-sugar cereals — higher income, more education, more time. ⚠️ Among the households with the highest intake of the product, it changed almost nothing.

⚠️ Which is the general pattern for education-only interventions: they work best on the people who needed them least, and they can widen the gap they were meant to close.

⚠️ Why it failed, in the terms this book has been building:

⚠️ It addressed knowledge when the constraint was not knowledge (Ch 32 §32.11). ⚠️ It required a decision at the moment of purchase, in an environment engineered against it (Ch 33 §33.4). ⚠️ It asked the individual to overcome a shelf position that had been purchased (§36.5b). ⚠️ And it is §36.7's bottom row — the weakest tier of the ladder, and the most politically comfortable, because it required nothing of anybody except the parent.

Intervention B — the reformulation and tax threat

⚠️ A decade later. A levy announced on high-sugar products in the category, with a threshold, a lead-in period before it took effect, and a lower rate for products below the threshold.

⚠️ The crucial design feature, and it is not the one people notice:

⚠️ The tax was announced BEFORE it applied, with a specific date and a specific threshold.

⚠️ Which gave manufacturers a period in which reformulating was cheaper than paying.

⚠️ What the industry said during the consultation:

⚠️ What actually happened to the product:

⚠️ Sugar content was reduced substantially, in stages, over about eighteen months. ⚠️ The reduction was achieved through a combination of reformulation and portion adjustment. ⚠️ Most consumers did not notice, and the ones who did adjusted.

⚠️ And the population-level result:

⚠️ Sugar purchased in the category fell — driven overwhelmingly by the change in what was IN the products rather than by any change in what people chose.

💡 The finding that matters. ⚠️ The intervention that changed what people ate did not require anybody to make a different decision.

⚠️ Intervention A asked millions of people to each make a better choice, repeatedly, forever. ⚠️ Intervention B changed one recipe.

⚠️ The comparison

⚠️ The inverse relationship in the last two rows is this case study's point.

⚠️ The interventions that work are the ones that are hardest to pass, and the ones that are easy to pass are the ones that do least. ⚠️ This is not a coincidence — an intervention that costs an industry nothing is not opposed by it.

⚠️ What Intervention B did NOT do

⚠️ Stated carefully, because this case study would be propaganda without it.

⚠️ The reformulated product is still a Group 4 product. ⚠️ Chapter 22's ward-trial finding was not attributed to sugar content, and a lower-sugar ultra-processed cereal remains an ultra-processed cereal.

⚠️ §36.5's both-halves point, demonstrated: reformulation genuinely works for the targeted nutrient and does not address what Chapter 22 identified.

⚠️ Two further limits:

⚠️ Substitution is possible. ⚠️ Reducing sugar can mean increasing other ingredients, and monitoring what replaced it is part of doing this properly.

⚠️ And a threshold creates a cliff. ⚠️ Products cluster just below it, which is a predictable response to any threshold-based rule and is an argument for graduated rather than binary designs.

⚠️ So: a real success, with real limits, achieved against sustained opposition, and worth understanding precisely rather than celebrating vaguely.

Analysis

1. ⚠️ Both interventions targeted the same product. Only one changed it.

2. ⚠️ The education campaign worked best on the households that needed it least, ⚠️ which is the recurring distributional failure of education-only approaches and the reason §36.7 ranks it bottom.

3. ⚠️ The levy's effect came through REFORMULATION, not through consumer response to price — ⚠️ which means the mechanism most people assume taxes work by is not the mechanism that did most of the work.

4. ⚠️ Announcing before applying is the design feature that produced that. ⚠️ A tax that arrives without warning collects revenue; a tax announced with a threshold and a date buys reformulation.

5. ⚠️ Every industry argument was made, and the technical-impossibility one was disproved by the industry itself within eighteen months. ⚠️ §36.13b's template, in one product.

6. ⚠️ The intervention that changed what people ate required nobody to make a different decision — ⚠️ Chapter 33 §33.4 at population scale, and §36.7's organizing principle.

7. ⚠️ And the political difficulty was inversely related to the effect. ⚠️ Which is worth generalizing: if a proposed food policy is uncontroversial, ask what it is not doing.

8. And it did not solve the Group 4 problem. ⚠️ A better cereal is not the same as a different breakfast.

Discussion Questions

1. ⚠️ The education campaign raised awareness and changed almost nothing. Is awareness worth anything on its own? ⚠️ What would you have to believe for the campaign to have been a good use of money?

2. ⚠️ It helped the households that needed it least. ⚠️ Find two other interventions in this book with the same distributional signature. (Chapters 30 and 32.)

3. ⚠️ Most of the levy's effect came from reformulation, not from price response. ⚠️ Does that make "tax" the wrong name for it? What should it be called, and does the name affect its politics?

4. ⚠️ "Reformulation is technically impossible" was disproved by the people who said it. ⚠️ How should a regulator treat that claim in future? How should a reader?

5. ⚠️ Political difficulty was inversely related to effect. ⚠️ Is that a general law? Construct a counter-example, or argue that you cannot.

6. ⚠️ The reformulated product is still Group 4. ⚠️ Is a better version of a problematic product progress or a distraction? (Ch 22 §22.9 has an argument on this.)

Your Turn

Step 1 — ⚠️ Find an intervention in your own country. A tax, a labelling rule, a marketing restriction, a school food standard.

⚠️ What it targets: __ ⚠️ Which §36.7 tier: __

Step 2 — ⚠️ Ask the diagnostic question.

⚠️ Does it require anybody to make a different decision at the point of purchase?

⚠️ If YES, expect a modest effect. If NO, expect a larger one.

Step 3 — ⚠️ Find the opposition arguments.

⚠️ Which of §36.13b's five templates were used? Nanny state · consumers will reject it · jobs · the science isn't settled · educate instead.

⚠️ Count them. It is usually all five.

Step 4 — ⚠️ Check the distributional effect.

⚠️ Who does it help most? __ ⚠️ If the answer is "people who were already doing the right thing," it is Intervention A.

Step 5 — ⚠️ And do the one thing §36.11's last row asks.

⚠️ Is there a live consultation, a school policy, or an institutional procurement decision you could respond to? __

⚠️ Twenty minutes. Almost nobody does it, which is precisely why it is worth doing.