Advance Layoff Notice Data from the WARN Act

We collect novel and timely data from advance layoff notices filed under the Worker Adjustment and Retraining Notification (WARN) Act. The act requires larger employers to notify affected workers at least 60 days before a potential mass layoff. We assemble WARN data from across the United States, and for many large states our data begin in the 1990s. We aggregate these data into an unbalanced, monthly panel of the state-level number of workers affected by WARN notices, and we update this panel twice a month. We also aggregate this panel to a national-level indicator of job loss (the "WARN factor") using a dynamic factor model.<br>

<b><br>Data Collection</b><br>We update the data twice a month by collecting WARN
notices from state websites. For many states, we have extended our historical
data by using digital archives of the internet and contacting state
officials.<br> <br><b>Citation</b><br>To learn more about
the data and the dynamic factor model, see:<br>Krolikowski, Pawel M. and Kurt G. Lunsford. 2022. “Advance Layoff Notices and Aggregate Job Loss.” Federal Reserve Bank of Cleveland, Working Paper no. 20-03R. https://doi.org/10.26509/frbc-wp-202003R.<br>Krolikowski, Pawel M., and Kurt G. Lunsford. 2024 “Advance Layoff Notices and Aggregate Job Loss.” Journal of Applied Econometrics. https://doi.org/10.1002/jae.3032.<b><br><br></b><b>File Description</b><br>The file labelled WARNFiles_YYYYMMDD.zip in the
main directory contains our most recent data. All data included in this zip
file were collected as of the date listed in the file name. Previous
vintages of our data appear in the `Archived_Vintages’ folder, with the same
naming convention.<br>
<br>The zip files contain three files:<br>
<br>

1.
README.txt

<br>2.
WARNData_NSA_YYYYMMDD.csv

This .csv contains the number of workers affected by
WARN notices by state and month. These data are not seasonally adjusted.
These data are the input into the dynamic factor model.

3.
WARNFactors_YYYYMMDD.csv

This .csv contains the output of the dynamic factor
model. The output is labeled as follows:

WARN Factor: the estimates of the factor from the
dynamic factor model.
MSE: the estimated mean squared errors
of the WARN factor.
WARN_hat: the number of workers affected by
WARN notices as implied by the WARN factor.
WARN_sum: the sum of the number of
workers affected by WARN notices from several states that form a
balanced panel beginning in January 2006. These states can change every
update.<b>Disclaimer</b><br>These data are
updated by the authors and are not an official product of the Federal Reserve
Bank of Cleveland.<br>

本研究基于《工人调整与再培训通知法案》（Worker Adjustment and Retraining Notification Act，简称WARN法案）归档的提前裁员通知，收集了新颖且时效性强的数据集。该法案要求大型雇主在潜在大规模裁员发生前至少60天，通知受影响的劳动者。我们从全美各地整合WARN法案相关数据，针对多数大州，数据集的时间跨度可追溯至20世纪90年代。我们将这些数据整合为非平衡月度面板数据集，统计各州受WARN通知影响的劳动者人数，且该面板每半月更新一次。此外，我们通过动态因子模型（dynamic factor model）将该面板数据整合为全国层面的失业指标——即“WARN因子”。

### 数据收集
我们通过抓取各州政府网站的WARN通知，每半月更新一次数据集。针对多个州，我们通过互联网数字档案及联络州政府官员的方式，补全了历史数据。

### 引用说明
如需了解数据集与动态因子模型的更多细节，请参阅：
Krolikowski, Pawel M. 与 Kurt G. Lunsford. 2022. 《提前裁员通知与总体失业规模》. 克利夫兰联邦储备银行，工作论文第20-03R号. https://doi.org/10.26509/frbc-wp-202003R.
Krolikowski, Pawel M. 与 Kurt G. Lunsford. 2024. 《提前裁员通知与总体失业规模》. 《应用计量经济学杂志》. https://doi.org/10.1002/jae.3032.

### 文件说明
主目录下名为`WARNFiles_YYYYMMDD.zip`的压缩文件包含最新版本的数据集。该压缩包内的所有数据均以文件名中标注的日期为截至采集时间。过往版本的数据集存放于`Archived_Vintages`文件夹中，命名规则与当前版本一致。

该压缩包内含三个文件：
1. README.txt：说明文档
2. WARNData_NSA_YYYYMMDD.csv：该CSV文件按州与月度统计了受WARN通知影响的劳动者人数，数据未经过季节性调整，是动态因子模型的输入数据集。
3. WARNFactors_YYYYMMDD.csv：该CSV文件包含动态因子模型的输出结果，各字段说明如下：
- WARN Factor：动态因子模型得到的因子估计值
- MSE：WARN因子的估计均方误差
- WARN_hat：由WARN因子推导得到的受WARN通知影响的劳动者人数
- WARN_sum：自2006年1月起形成平衡面板的若干州的受影响劳动者人数总和，该州组会随每次更新发生变化。

### 免责声明
本数据集由作者自行更新，并非克利夫兰联邦储备银行的官方产品。