Doubts on input quality: The effect of inaccurate fertilizer content on the estimation of production functions and technical efficiency

Household survey responses regarding levels of input use are sometimes affected by bias of which even the households themselves are not aware. Some examples include poor quality seed with a mixture of the fertile and infertile types, and pesticide content that has been substituted with less effective chemicals. We analyze in our paper the effect of low quality fertilizer, which contains less nitrogen than is advertised on the packaging. We show that this could lead to bias in the estimation of marginal effect and technical efficiency. Using panel data from the Hebei province of China, we calculate the magnitude of the bias across different levels of fertilizer quality. We find that the bias could be between -2% and -7% for marginal effect of fertilizer at mean input levels, and between 1% and 4% for technical efficiency.