Regulatory capture in a resource boom

Abstract

States oversee most regulation of oil and gas extraction in the United States. When relying on regulation, state oil and gas agencies may be susceptible to capture by firms developing resources. This may be particularly problematic during booms of resource development when information asymmetries are largest and existing regulations risk becoming obsolete. If regulators are captured, they may take actions that serve concentrated private interests in preference to the public interests they are charged with upholding. I develop and test hypotheses that oil and gas regulators are captured. The primary empirical tests use data from state regulation in North Dakota to prevent resource waste by restricting natural gas flaring. The empirical results are consistent with the theory of regulatory capture, providing empirical evidence that captured regulators serve well-organized specific interests in preference to diffuse general interests. These results provide novel granular evidence of the mechanisms for regulatory capture by showing differences in regulatory responses across firms and locations. This detailed evidence has implications for the design of regulations and reliance on regulatory interventions to protect the public interest.

Appendices

Appendix

A model of flaring

The first key feature of natural gas flaring in North Dakota is that oil and natural gas are co-produced. Separable investments in gathering and processing capacity are necessary to collect and market oil and natural gas. If these investments are not made, then the gas must be flared thanks to the lack of alternatives. The opportunity cost of flaring is the foregone net benefit of capturing and using the gas. The benefit of flaring is the time value of avoiding a delay in the net value of oil produced.

2.1 Production dynamics

Oil and natural gas production is dynamic. Consider a simple model of the dynamics in which two products decline at the same constant rate. Consider the production of two outputs $y^1$ and $y^2$ that are co-produced over time. For each period t:

$$\begin{aligned} {\textbf{y}}_t = \left( 1 - \alpha \right) {\textbf{y}}_{t-1} \end{aligned}$$

(A.1)

which allows us to calculate the ultimate recovery of product i as a function of initial production— $\frac{{\textbf{y}}_0}{\alpha }$ over an infinite horizon. Cumulative production at any time $\tau$ is $\mathbf {y_0} \sum _{t=0}^{\tau } \left( 1 - \alpha \right) ^t$. Estimable parameter $\alpha$ is interpreted as the average decline rate.

An essential fixed investment $D>0$ allows joint production of two outputs, $y_1$ and $y_2$, which are produced in fixed proportion $\xi$. After investment D is made, the costs of collecting and marketing the outputs are separable. These costs have both fixed and variable components. Normalize the fixed cost of collecting and marketing $y_1$ to zero so that the fixed costs of collecting and marketing $y_2$ are always higher. The firm has two potentially profitable options, depending on which investment option is chosen.¹⁹

$$\begin{aligned} \Pi = {\left\{ \begin{array}{ll} \left( p_1 - c_1 \right) y_1 &{} I_1=1, I_2=0 \\ \left( p_1 - c_1 \right) y_1 + \left( p_2 - c_2 \right) \frac{y_1}{\xi } &{} I_1=1, I_2=1 \\ \end{array}\right. } \end{aligned}$$

(A.2)

In the first case, $y_1$ is produced and $y_2$ is freely disposed. In the second case, both products are sold. The choice of whether to pursue free disposal as a strategy depends on the total profits associated with each strategy. Production occurs over time, so the decision to invest at time $\tau$ is a function of the difference in profits. In the simplest setting, this decision rule requires that the net present value of $y_2$ must exceed the fixed investment required to collect and market it, taking the time value of money into account via discount factor $\beta$.

$$\begin{aligned} B_t(I_2; y_{1t} ) = \sum _{t=\tau }^T \left\{ \beta ^t \left( p_{2t} - c_{2t} \right) \frac{y_{1t}}{\xi } \right\} - I_2 \gtrless 0 \end{aligned}$$

(A.3)

This is comparable to previous results when the firm’s investment decision is isolated from spillovers on other firms.²⁰

The expected benefits of connection are unlikely to be observed directly. Suppose that the probability that an investment is made to instantaneously connect the well is a monotonically increasing function of the expected gains from connection.

$$\begin{aligned} \textrm{Pr}(I_2=1) = F(B) \end{aligned}$$

(A.4)

The function F satisfies the conditions of a cumulative distribution function.

Comparative statistics on (A.4) yield three empirical predictions. Assuming that well operators are profit maximizers, any increase in the net price of captured gas should decrease flaring. A higher output price makes the flared gas worth more and thus makes connection to the marketing network more likely. Similarly, increasing the cost of free disposal (such as by requiring higher payments for taxes and foregone royalties on flared gas) should increase the probability of connection. Conversely, if the required fixed investment to make a connection is higher, we would expect that operators are less likely to decide to connect.

2.2 Revenue variability

The dynamic nature of the problem also introduces the prospect of $p_2$ varying over time. Changing the market price for natural gas will likely alter the net benefits from reducing flaring. Suppose that the direction of change in $p_2$ in any discrete time interval $\Delta t$ is determined by a binomial distribution: the probability of an upward price movement of magnitude $\delta$ is $\rho$, and the probability of downward price movement $\frac{1}{\delta }$ is $1-\rho$. This allows for a recombining lattice as described in Cox et al. (1979), which approaches a continuous Weiner process as $\Delta t$ becomes short. Figure 6 depicts the possible paths for price evolution over time periods. The connection investment is effectively irreversible and specific to a particular well, or group of wells in the case of a gas plant. However, this investment option is different from many similar problems insofar as production declines over time subject to A.1.

The deterministic physical decline is coupled with stochastic prices to modify the left-hand side of (A.3). Revenues in period t are $y_{t-1}(1 - \alpha ) p_t \delta$ if price increases and as $\frac{y_{t-1}(1 - \alpha ) p_t}{\delta }$ if price decreases. The straightforward specification allows us to combine price volatility and physical decline to a single measure of revenue volatility. This slightly adjusts the values of the underlying asset at the nodes of Fig. 6. Assuming that the decline rate is positive, this makes waiting to invest in gathering less attractive, and therefore should decrease the value associated with waiting as time progresses.²¹

Fig. 6

Binomial price tree for depleting asset

For a given initial production rate $y_0$, each node can be assigned an expected future revenue and an option value associated with waiting to connect, conditional on expectations about volatility. The expected future revenue can be compared with the investment cost. The investment cost may change over time, for example as an existing gathering network is installed closer to a given well. Represent the option to connect to gathering as a function of the age of the well, the current price level and expected volatility, the current investment cost needed for connection, and the uncertainty about the continuation value of the option.²² The value of the option varies with the predetermined initial production level, the underlying remaining revenue (A), and the investment cost ($I_2$).

$$\begin{aligned} V(A(t), p_t, I; y_1, T) = {\left\{ \begin{array}{ll} \sum _{t=\tau }^T \left\{ \beta ^t \left( \overline{p_{2t} - c_{2t}} \right) \frac{y_t^1}{\xi } \right\} - I_2 &{} \textrm{Exercise }\\ e^{-r}V(A(t+1), p_{t+1}, I_2; y_1, T) &{} \textrm{Cntinuation} \end{array}\right. } \end{aligned}$$

(A.5)

The value of the option is strictly decreasing in the underlying revenue A and I, and increasing in p. The option should be exercised most often for wells that are relatively early in their productive lives, at times when the price is relatively high, and when connection costs are relatively low. The option has a term of the expected economic life, T, but because production deterministically decreases over the life of the well, it is likely to fall in value over time.