Why hospitality spending is falling when households have €1,057 more to spend

€236 more per household on food for home. €92 less on eating and sleeping out.

Spaniards have not stopped spending on food. They have decided where they eat it.

And this is not a bad year. It is a reallocation of the household budget, much slower to reverse than a bad season. The headline that hospitality spending is "freezing" undersells it: it is not freezing, it is falling, and it falls in the year everything else rises.

This same week another headline said the opposite: that hospitality is already Spanish households’ top spending item.

And it is true, according to Eurostat. Restaurants and lodging are 18.1% of household spending in Spain in 2026, ahead of food at 16.9%.

INE says 9.3%.

Nine points between two official statistics with the same name. INE surveys households that live here; the Eurostat series captures spending in the territory, tourists included.

When you read that hospitality is Spain’s consumption engine, they are talking about the tourist. Your neighbour is in the 9.3% column, and that column is falling.

If your till depends on locals, that is your figure. We broke down the tourist side in what tourist spending really left this July, and that is not what it looks like either.

First the part that hurts. What has fallen is not food: it is drinks. According to the Ministry of Agriculture out-of-home consumption panel in Q1 2026, reported by Hostelería Madrid, food grew 1.2% in volume and drinks fell 3.4%.

Beer accounts for 94.1% of the entire drinks loss.

Customers still eat out, but they have dropped the beer afterwards, Saturday’s aperitif and the second round. Exactly the highest-margin part of your menu.

And now the good news, which is also real: out-of-home consumption penetration rose 1.3 points, to 91.4%. What is falling is frequency, 3.1% fewer occasions, and consumptions per occasion, 4.4% fewer.

You have not lost customers. You have lost visits and rounds.

Spain has more than 19.3 million households, according to INE projections, and more than 300,000 hospitality venues, according to the Spanish Hospitality Yearbook.

€92 less per household is about €1,776 million that left the sector in a year. Split across every venue, €5,900 less a year each, about €490 a month. Tipi’s own linear split over INE figures.

Two honest caveats. Lodging is in that group too, so bars and restaurants get less. And those €490 are sales, not margin: at the 3% or 4% net margin associations talk about today, that is about €15 or €20 of profit a month.

Nobody closes for losing €490 of sales a month. They close for losing it twenty-four months in a row without seeing it in any figure.

Tipi’s own estimate with 33% food cost and 22% drink cost, plus €18 to serve the table in labour and venue cost. Not a market figure.

A table for two, a Friday, ninety minutes.

Sales fall 45%. What is left after paying for that table falls 79%.

Because the table takes the same space, the waiter costs the same and the venue costs the same. What has fallen is drink, which on that bill was 39% of the ticket and 42% of the margin.

That is why full dining rooms can still lose money. Covers have not fallen. The mix of the bill has.

1. Split your sales into food and drink, month by month, for the last two years. If drink has lost share, you have found the hole and it is not footfall.
2. Look at tickets by daypart. If lunch holds and night falls, the same thing is happening to you as to the panel.
3. Calculate margin in euros, not percentage, on your five best-selling drinks. The percentage barely moves when sales fall; the euros do.
4. Cost the dishes that gained share this year before you print the winter menu. The customer chose them, not you.

The HBS is for 2025 and was published in June 2026. It is the most reliable snapshot of the household budget, but it is not this month’s. If 2026 reverses the trend, we will not know until June 2027.

Eurostat’s 18.1% and INE’s 9.3% both stand, and both are official. Our reading is that the gap is tourists and lodging, but neither source spells it out that way. If a breakdown appears, we will update this.

Q1 2026 panel data is a sample, not a census. It marks direction with more confidence than magnitude.