Online Program

The contribution of labor reallocation to aggregate productivity growth: a synthetic data approach for the United States
Benoit Dostie, HEC Montréal 
*Lars Vilhuber, Cornell University 


Keywords: Synthetic data, longitudinal, business register, administrative data, confidentiality protection, imputation

A large literature attempts to quantify how factor reallocation contributes to productivity growth. Prominent papers using a variety of methods include Bartelsman and Doms (2000), Foster, Haltiwanger and Krizan (2000), and Lentz and Mortensen (2009). We present results from an extension of the Lentz and Mortensen (2009) model to the US data (and a re-estimation of FHK results with more recent data), assessing the sensitivity of results to the length of the data series and cyclical patterns in the data. We use both the Synthetic and the confidential LBD, and describe results from an analytical validity assessment of the synthetic data. Since the synthetic data do not contain all the data moments necessary to estimate the LM model, we further describe a simple complementary mechanism to combine synthetic data and partially synthetic moments.