Working Paper: NBER ID: w1733
Authors: Steven G. Allen
Abstract: This paper estimates the effect of union work rules in the building trades on employment and costs by comparing factor demand elasticities for union and nonunion contractors and subcontractors over micro data from two different types of construction. The results show that the elasticities of substitution between labor and nonlabor inputs and own-price elasticities for nonlabor inputs are about the same for union and nonunion contractors. In contrast, the elasticities of substitution among different skill categories of labor and the own-price elasticities for each category are much lower under unionism. A simulation based on a typical office building subcontract shows that these lower factor demand elasticities result in excess staffing of 3.2 percent, excess labor costs of 5.0 percent, and excess total costs of 2.0 percent. This study also examines directly the effect of union work rules on the use of prefabricated components and finds that union contractors are justas likely to use them as nonunion contractors.
Keywords: Union work rules; Factor demand elasticities; Construction efficiency; Labor markets
JEL Codes: J51; J53; L74
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| Union work rules (J51) | Lower elasticities of substitution among different skill categories of labor (J24) |
| Union work rules (J51) | Lower own-price elasticities for each category compared to nonunion contractors (L79) |
| Lower elasticities of substitution (D11) | Excess staffing of 32% (J45) |
| Lower own-price elasticities (D40) | Excess labor costs of 50% (J39) |
| Lower elasticities of substitution (D11) | Excess total costs of 20% (D24) |
| Union work rules (J51) | Restriction of managers' flexibility to adjust input quantities in response to price differentials (L11) |
| Lower elasticities (H30) | Influence on overall costs and productivity (D24) |
| Union work rules (J51) | No significant difference in likelihood of using prefabricated components (L74) |