UC Berkeley

This dissertation consists of three chapters. The first chapter investigates whether U.S. oil market policy succeeded in lowering the price of crude oil. The second chapter examines the effect of private sponsorship of university research on the allocation of expenditures between public good research and commercial applications. The third chapter estimates the effects of congestion pricing in the Bay Area on public transportation ridership and air pollution levels.

The first chapter, ``The Strategic Petroleum Reserve and Crude Oil Prices'', uses a structural vector autoregression (VAR) model of the U.S. oil market to estimate the effect of the Strategic Petroleum Reserve (SPR) on crude oil prices.

In contrast to existing literature, I show that unanticipated oil releases from the SPR have no measurable effect on oil prices. However, unanticipated oil purchases for the SPR raise oil prices 1.5 percent.

I confirm these estimates with two alternate VAR identification strategies that use external information, rather than the traditional recursiveness assumption.

First, the SPR purchase schedule is used as an instrument in the VAR model to identify the effect of SPR purchase shocks on oil prices. I verify the statistically significant purchase effect and show that standard identifying assumptions have a downward bias.

Second, I identify SPR release shocks with high-frequency futures market data. Using several futures contracts to estimate unanticipated SPR policy shocks, I find that SPR releases do not have an effect on oil prices.

To explain this puzzle, I estimate an interacted VAR with time-varying policy effects. This model shows that SPR purchases only increase oil prices when uncertainty is high, but SPR releases do not lower oil prices at any level of uncertainty.

The second chapter, ``Public vs. Private Good Research at Land Grant Universities'', examines the effect of private sponsorship of university research on the allocation of expenditures between public good research and commercial applications. Throughout the land-grant university system, there is much concern that as a result of reduced government funding, fundamental research will be neglected at the expense of research that is geared toward commercial applications. This paper attempts to shed some light on the relationship between research priorities and the availability of public funding for university research. In particular, we use both a static and a dynamic model to investigate the conditions under which university/private research partnerships can crowd-in or crowd-out basic science research as public funding becomes scarcer.

In the third chapter, ``The Effects of Congestion Pricing on Public Transportation Demand and Air Pollution'', I examine the effects of congestion pricing in the Bay Area. On July 1, 2010, congestion pricing was implemented on the San Francisco-Oakland Bay Bridge. Using panel data on hourly commuter rail (BART) and commuter bus (AC-Transit) ridership from June 1, 2009 through July 31, 2011, we examine the effect of congestion pricing on public transportation usage and air pollution. Using a difference-in-difference model to control for unobserved variables, we estimate that BART ridership rose 4-8 percent during peak hours, while AC-Transit ridership rose 11-15 percent. Our estimated cross-price elasticities for public transportation, 0.2 for commuter rail and 0.15 for commuter bus, are slightly below estimates from the transportation economics literature. We also use panel data on air pollution levels in the Bay Area during the same time period to examine the effect of congestion pricing on air pollution. Using a similar difference in difference model, we find air pollution levels are flat during peak hours, but rise during off peak hours. As a robustness check, we use regression discontinuity models, which suggest similar, but slightly smaller effects of congestion pricing on air pollution and public transportation.