NBER Publications by Ryan Kellogg

Researchers estimating the demand for energy-using durable goods must specify consumers' beliefs about future energy prices. Policy-relevant inference hinges on this specification, yet there is little direct evidence on the nature of consumer beliefs. We provide such evidence by analyzing two decades of data on gasoline price expectations from the Michigan Survey of Consumers. We find that average consumer beliefs are indistinguishable from a no-change forecast. This finding has important implications for the literature on consumer valuation of energy efficiency, and it implies that researchers are likely justified in assuming a no-change forecast, as is common practice.

Despite widespread application of real options theory in the literature, the extent to which firms actually delay irreversible investments following an increase in the uncertainty of their environment is not empirically well-known. This paper estimates firms’ responsiveness to changes in uncertainty using detailed data on oil well drilling in Texas and expectations of future oil price volatility derived from the NYMEX futures options market. Using a dynamic model of firms’ investment problem, I find that oil companies respond to changes in expected price volatility by adjusting their drilling activity by a magnitude consistent with the optimal response prescribed by theory.

This paper examines the importance of learning-by-doing that is specific not just to individual firms, but to pairs of firms working together in a contracting relationship. Using new, detailed data from the oil and gas industry, I find that the joint productivity of an oil production company and its drilling contractor is enhanced significantly as they accumulate experience working together. This learning is relationship-specific: drilling rigs generally cannot fully appropriate the productivity gains acquired through experience with one production company to their work for another. This result is robust to other ex ante match specificities. Relationship-specific learning is consequential because it implies that relationship stability is important to productivity. When two firms accumul...

Regulators and firms often use incentive schemes to attract skillful agents and to induce them to put forth effort in pursuit of the principals' goals. Incentive schemes that reward skill and effort, however, may also punish agents for adverse outcomes beyond their control. As a result, such schemes may induce inefficient behavior, as agents try to avoid actions that might make it easier to directly associate a bad outcome with their decisions. In this paper, we study how such caution on the part of individual agents may lead to inefficient market outcomes, focusing on the context of natural gas procurement by regulated public utilities. We posit that a regulated natural gas distribution company may, due to regulatory incentives, engage in excessively cautious behavior by foregoing surplus...