Curated Measures

Curated Measures is the NL Analytics library of maintained curated datasets for established research constructs.

What Curated Measures contains

Each curated dataset is built on the same earnings-call corpus as your datasets. Its entry explains the source research, the measures it contains, and when its living dataset was last refreshed.

Country Risk is the currently approved named example. It contains per-country measures based on research published in The Review of Economic Studies. Its headline aggregate time series is the NLA Country Risk Index.

Open a curated dataset to see its description, source research, measures, and access details.

Living data and vintages

The living dataset follows corpus refreshes and may be newer than a frozen vintage.

When to use a curated dataset

Use a curated dataset when the construct is established in the literature and the maintained method fits the research design.

Create your own dataset when the construct is novel, needs different terms or exclusions, or requires your own validation. A curated dataset can also serve as a benchmark for the measures in your dataset.

The political-risk and country-risk families are also available through weighted dictionary queries, which use the underlying weighted dictionaries in the regular search workflow.

Interpret curated datasets carefully

Curated datasets are text-derived. Their measures capture discussion in earnings calls, not real-world outcomes. The caveats in What the measures do not mean apply. Cite the source paper named in the curated dataset entry and record the living dataset's refresh date with the export.