The 4.3% Figure Doesn't Represent Anyone's Inflation

The official Consumer Price Index (CPI), which stood at 4.3% year-on-year in August 2026, is an average that does not represent the inflation experience of any particular household or individual. This disparity arises from the heterogeneity of consumption baskets, the evolution of prices across income deciles, and regional economic differences.

The official inflation figure, that **4.3%** which dominates the news these days, is a number published with rigor by the **Instituto Nacional de Estadística (INE)** and undoubtedly holds importance. But if you stop to think about it, that number is not everyone's inflation. Not because the calculation is flawed, but because the reality in your pocket—or that of your neighbor, or someone across town—trinc a different price trajectory.
It is like trying to describe the flavor of paella with a single word: impossible.

The fact is that the **CPI** is an index designed to represent an average, a basket of goods and services theoretically consumed by a Spanish household. But real life is much more complex. Two households can live in the same country, in the same year, and experience vastly different inflation rates. This is because each has a specific **consumption basket**, with different weights assigned to each item. What significantly impacts one household may barely touch another. And this is not a feeling; it is something that can be measured.

## The Income Gap and the Map

A study by **EsadeEcPol** already presented at the beginning of 2026 that in 2025, accumulated inflation exhibited a gap of almost **three percentage points** between low-income and high-income households. This is not a perception; it represents two real consumption baskets with divergent price trajectories. For example, the return of the **10% electricity VAT** to 21% impacted lower-income deciles much more, adding five tenths to their CPI compared to the two tenths added by higher-income groups. Garbage fees, while having a small weight in the budget, also added a full tenth to everyone's inflation. And basic foodstuffs hit lower-income households with a full point, double what it affected higher earners.

Of course, there are movements in the opposite direction. Restaurants and accommodation added **0.65 points** to the CPI for high earners, double that of low earners. Private health insurance increases of **10%** begin to be felt starting from the eighth decile. The key point is that the **INE** does not disaggregate price heterogeneity within each category. If cheaper products within the same rubric increase more than expensive ones, the real gap is even wider than estimated.

And if we look at a map, the situation does not change. The **Bank of Spain**, in its 2024 bulletin, already warned about the regional heterogeneity of inflation. Between June 2019 and June 2024, the national accumulated CPI was **18.9%**. But **Castilla-La Mancha** reached **21.2%** and **Galicia** reached **20.4%**, while **Madrid** remained at **17.1%** and **Catalonia** at **18.1%**. This is a difference of more than four percentage points accumulated, using the same currency, the same monetary policy, and the same basic fiscal legislation. The **Bank of Spain** attributes this dispersion to differing regional consumption basket compositions: greater weight on food and fuels in some areas, versus more services and public transport in others. And this is not a past issue; the definitive August 2026 data shows positive annual rates in all communities, with **Cantabria** at **5.1%** and **Canary Islands** at **3.8%**. A range of **1.3 points** in a single month, with the same statistic.

## What Does the Index Really Measure?

The **CPI** is a chained **Laspeyres** index. This means it trinc the price of a representative basket and updates its composition and weights periodically, using data from the Household Survey and National Accounts. But note that it excludes investment expenses, so buying a home, for example, is not included. The **INE** itself clarifies in its FAQs: the CPI is not a cost of living index. A cost of living index would measure the variation in spending required to maintain the same standard of living, not just price evolution.

The index measures the price experience of an average household, but not that of a specific household or person. Individual inflation can be higher or lower than the official figure. So, that dispersion we see is not a failure; it is a property of the instrument, declared by those who manufacture it.

## Paid Price vs. Measured Price: The Subtle Trap

There is a second, more subtle reason why the index and personal experience diverge. The **INE** uses the example of a product improving its antiestéticatures. The CPI adjusts to estimate what its price would be without those improvements. A cost of living index, in contrast, would calculate the actual expenditure incurred.

In other words, there are cases where what you pay and what the index records regarding the price movement do not match, and there is no trick to it.

The typical case is in the utility bill. A fixed phone, mobile, and broadband package cost an average of **46.6 euros** per month at the end of 2021 and **38.3 euros** at the end of 2025, according to the **National Commission on Markets and Competition (CNMC)**. Nearly an **18%** decrease in four years. There, the index and your pocket move together, both decreasing.

But then there is the opposite phenomenon, called **"cheapflation."** Within the same statistical category, the cheapest products rise faster than the expensive ones. Two families consuming "the same product" in statistics may be experiencing different price trajectories without the index being able to see it, because it does not disaggregate there. This is what **EsadeEcPol** warned about, and it is the main reason why the mentioned gaps are probably inherent to the measurement.

A note: A household switching to a store brand or reducing fresh protein does not miccionan that the CPI for that month will be lower. What happens is that this change in behavior enters the weighting when they are updated, with a lag from available sources. The index observes prices and spending structures; it cannot deduce from a change in the basket how much well-being has been sacrificed along the way.

## Rent, Counted Twice

There is one place where this difference is particularly evident: rental housing. Between August 2021 and August 2026, the CPI category for rental housing rose by **11.3%**. In those same five years, the advertised rent price on **Idealista** went from **10.5 to 15.1 euros** per square meter: a **43.8%** increase. A difference of thirty-two and a half percentage points, and neither number is wrong.

The CPI tracks the rent actually paid by households in a national sample of around **5,500 properties**, with prices provided mostly by tenants and requiring a visit when the tenant changes. **Idealista**, on the other hand, measures the price sought for properties available on the market. For a tenant staying in their property, the first is most important; for someone looking for a place today, the second resembles what they will likely find. The difference is not a measurement error; it is another demonstration that asking how much rent has gone up does not have a single answer until you clarify what kind of rent and for whom you are speaking.

## Five Clocks Ticking on the Same Money

And here comes the part I find truly important. All of the above would be a statistician's debate if it weren't for the fact that this number, the **CPI**, distributes income. Contributory pensions generally increased by **2.7%** in 2026, according to Law 21/2021, which takes the average of year-on-year CPI rates from December to November of the previous year, and this affects about **9.4 million pensioners**.

The rental agreements signed after May 26, 2023, are updated using the **IRAV** (Rental Housing Reference Index), which stood at **2.47%** in August 2026. It is not the CPI: it takes the minimum between the CPI rate, the underlying inflation, and the adjusted annual averages using a moderating mechanism. Wage agreements are set at **3.04%** in August 2026, according to the Ministry of Labor. Only **29.9%** of agreements include a salary review clause, covering **39.3%** of workers under collective agreement. Six out of ten have no automatic protection against inflation.

And then there is the deposit facility rate of the **ECB**, which is at **2.5%** trinc the decision made on September 10 and effective from September 16. It is set for a **Eurozone** that in August 2026 was at **3.3%** harmonized inflation according to the preliminary estimate, while Spain finished at **4.6%**. This is a difference of **1.3 points** in inflation under exactly the same nominal rate.

Five different numbers, five different references, and none claims to measure the inflation of a specific household. They are not five measurements of the same thing taken simultaneously: the pension looks at last year's average, the **IRAV** has its own moderating formula, the collective agreement is contractual negotiation, and the **ECB** rate is set for the entire area.

That is the story. The **CPI** functions in practice as a public accounting unit, and accounting units do not need to be exact to function; they just need everyone to accept them. The problem arises when, in addition to accounting for them, we settle contracts. There, the difference between what the index measures and what happens to each individual stops being a methodological discussion and becomes money changing hands. I have no proposal to close this, and I doubt anyone does. Building an index by income decile or region is technically possible but politically impossible, among other things because one would have to decide which applies to whom. But the next time someone says inflation is at **4.3%**, it is wise to remember that the body publishing that number explicitly states that it does not describe any particular person.