1. Standard errors are in parenthesis. The coefficient of interest is the effect of IT outsourcing on productivity. We use the "Y2K bug consulting" as the IV for IT outsourcing. Columns 1-3 show the results of the fully parametric model, using two-stage least squares and instrumenting IT outsourcing using the Y2K bug. The specifications in Columns 4-6 nonparametrically control for investment and firm size (number of employees), and parametrically for the share of employees working predominantly with personal computers, sector, location in Eastern Germany, and whether the company indicated interest in the survey results. Columns 4-6 use non-parametric Kernel estimation. Columns 1 and 4 show the biased benchmark using listwise deletion (MCAR). Columns 2 and 5 correct for missingness using imputation methods that assume missing at random (MAR). Columns 3 and 6 apply the correction for selective non-response that we;Notes: This table shows our main estimation results for the effect of IT outsourcing on productivity. The dependent variable is log(sales-costs of intermediates/employees)
2. Firms' use of outside contractors: Theory and evidence;K G Abraham;Journal of Labor Economics,1996
3. Trade liberalization, intermediate inputs, and productivity: Evidence from Indonesia;M Amiti;The American Economic Review,2007
4. Service offshoring and productivity: Evidence from the United States;M Amiti;The World Economy,2009
5. Productivity effects of organizational change: microeconometric evidence;I Bertschek;Management Science,2004