Firm Capacity Underutilization and the Measurement of Productivity

When estimating a production function, economists usually assume that firms fully employ all their available inputs. Contrary to this assumption, we document that underemployment of quasi-fixed inputs or low capacity utilization rates (CUR) is common across firms, especially in poor countries. Low CURs are correlated with supply-side constraints such as shortages of material inputs and electricity; they are also correlated with demand-side constraints such as a lack of access to markets. We show that when firms underemploy their quasi-fixed inputs, the assumptions behind standard production function estimation techniques-such as the control function approach-are invalid, and these techniques produce biased estimates of TFP. We demonstrate this using unique panel data from Ethiopia with information on actual and full capacity input demand. We find that measures of TFP that do not account for CUR considerably underestimate ’true’ productivity when CUR is low but not when CUR is high. This leads to an exaggeration of the TFP gap between rich and poor countries.