# Cost of residential space cooling by country under climate and adaptation scenarios

**Dataset:** `cost_cooling_bycountry_scenario_*` (6 files)
**Author:** Giacomo Falchetta, CMCC Foundation (RFF-CMCC EIEE), `giacomo.falchetta@cmcc.it`
**Produced within:** ACCREU (Assessing Climate Change Risk in EUrope), Horizon Europe
**Version:** outputs of 10 April 2026

- **Sector:** residential only. Services and industry are not covered here. They could in principle
  be approximated by upscaling with the top-down sectoral projections, but that is not done in these
  files.
- **Spatial:** 133 countries in the files as delivered. *[TO CONFIRM: the April 2026 correspondence
  refers to 179 countries based on the shapefile used. The delivered files contain 133 distinct
  ISO3 codes. Please confirm which is correct for the deposited version.]* Countries absent from the
  files are outside the workflow's domain, not zero-cost.
- **Temporal:** 2020-2100, annual, cumulative from a 2020 base of zero.

---

## 1. Summary

Bottom-up projections of the cost of space cooling in the **residential sector**, by country and
year, under three ACCREU adaptation intensiveness variants, three climate scenarios, and a
no-climate-change counterfactual for each.

The estimates cover upfront investment in the air-conditioning (AC) stock (the extensive margin)
together with the electricity expenditure for operating it (the intensive margin), reported as four
separate cost components so that users can select the ones relevant to their application.

The projections are derived from a bottom-up cooling-cost model built on a non-linear,
non-parametric machine-learning estimate of AC ownership and cooling needs as a function of cooling
and heating degree days and socioeconomic drivers (Falchetta et al. 2024). 

Each `value` is the cost accumulated **from 2020 up to and including the year in the `yr` field**.
The series are monotonically non-decreasing by construction and start at zero in 2020. To obtain the **annual (incremental) cost** in a given year, take the first difference against the
previous year, within each country / component / scenario / forcing-level group. 

If these data are combined with the ACCREU top-down energy-demand datasets, drop the
`Electricity` component. Those datasets already report energy expenditure for the residential
sector, in addition to the other sectors, so retaining `Electricity` here would count the intensive
margin twice. Use the three investment components from this dataset for the extensive margin and
the top-down datasets for the intensive margin.

## 3. Files

| File | Adaptation intensity | Climate |
|---|---|---|
| `cost_cooling_bycountry_scenario_base.csv` | base | with climate change |
| `cost_cooling_bycountry_scenario_mid.csv` | mid | with climate change |
| `cost_cooling_bycountry_scenario_high.csv` | high | with climate change |
| `cost_cooling_bycountry_scenario_base_nocc.csv` | base | no climate change |
| `cost_cooling_bycountry_scenario_mid_nocc.csv` | mid | no climate change |
| `cost_cooling_bycountry_scenario_high_nocc.csv` | high | no climate change |


## 4. Variables

Long (tidy) format, comma-separated, UTF-8, quoted character fields, no missing values.

| Column | Type | Description |
|---|---|---|
| `yr` | integer | Year, 2020 to 2100, annual resolution. |
| `variable` | character | Cost component.
| `value` | numeric | **Cumulative** cost since 2020, in **2021 US dollars at market exchange rates** (not PPP, not millions or billions: the unit is single dollars). Undiscounted. |
| `ISO3` | character | ISO 3166-1 alpha-3 country code, from the shapefile used in the underlying workflow. |
| `SSP` | character | Climate scenario: `26`, `45`, `70` for **SSP2-RCP2.6, SSP2-RCP4.5, SSP2-RCP7.0**, per the ACCREU protocol. Socioeconomic assumptions are SSP2 throughout; only the forcing level varies. |

### Cost components

The four categories are **mutually exclusive and additive**: their sum is total cumulative cooling
cost for that country, year and scenario.

| Value (exact string, incl. leading spaces) | Type | Description |
|---|---|---|
| `"   Increasing penetration"` | investment | Purchase of AC equipment in dwellings entering the cooled stock, i.e. the cost of extending AC access. |
| `" Retrofitting"` | investment | Refurbishment and replacement of equipment in the pre-existing cooled stock. |
| `"  Retrofitting of new penetration"` | investment | Refurbishment and replacement associated with the newly cooled stock. |
| `"  Electricity"` | expenditure | Cost of the electricity consumed for space cooling. |

## 5. Scenario dimensions

### Adaptation intensity: `base`, `mid`, `high`

The three ACCREU adaptation intensiveness variants, defined in **Deliverable D2.2**. Costs increase
monotonically with intensity (`base` < `mid` < `high`).

### Climate: main files vs `_nocc`

`_nocc` files are a **no-climate-change counterfactual**: socioeconomic drivers (population, income,
urbanisation) evolve along SSP2 while the climate signal is removed.

Subtracting the `_nocc` file from its matching main file, for the same adaptation variant,
component, country and year, gives the **cost attributable to climate change**, net of
socioeconomic drivers:

```
climate-attributable cost = base - base_nocc     (and likewise for mid, high)
```

## 6. Using the data

Aggregating countries to model regions is a straightforward **sum**, since values are levels rather
than percentage changes, and no weighting is required.

## 7. Contact

Giacomo Falchetta, CMCC Foundation, `giacomo.falchetta@cmcc.it`