Economic bene…ts of management reform in the northern

Gulf of Mexico reef …sh …shery

Quinn Weninger and James R. Waters¤

January 2002

Abstract

Controlled access management in the northern Gulf of Mexico commercial reef …sh…shery has failed to achieve economic and biological objectives. This paper estimatesthe economic bene…ts of replacing controlled access with a system of tradable harvestpermits. Using 1993 data, we estimate that eliminating market gluts caused by periodicseasonal closures could increase revenues by $3.068m. Harvest cost savings from elimi-nating per-trip catch limits and periodic seasonal closures are estimated to be $8.547m.Total 1993 bene…ts are estimated at $11.614m suggesting a property rights-based man-agement program is an attractive alternative in the northern Gulf reef …sh …shery.

Key Words: Rights-based management reform, harvest e¢ciency, rent generation.

JEL Classi…cation: Q2, D2

¤Do not cite without author permission. Authors are respectively, Assistant Professor, Department ofEconomics, Iowa State University, and Economist, US Department of Commerce, National Marine FisheriesService. We thank seminar participants at Iowa State University and the University of Minnesota. Specialthanks to Felix Cox, Mike Travis and Peter Emerson. We alone, of course, are responsible for any remainingerrors and omissions. Potential management implications of this paper do not necessarily re‡ect the manage-ment philosophy of the National Marine Fisheries Service. Please send correspondence to the Department ofEconomics, 260 Heady Hall, Iowa State University, Ames, IA, 50011-1070, or email weninger@iastate.edu.

1

1 Introduction

Controlled access …sheries management, consisting of vessel entry restrictions, a total allow-

able catch (TAC) policy that is enforced with periodic …shery closures, and per-trip catch

limits for qualifying vessels, has failed to achieve biological and economic management ob-

jectives in the northern Gulf of Mexico commercial reef …sh …shery. The National Marine

Fisheries Service has classi…ed the stock of the highest valued species in the northern reef

…sh complex, red snapper, and 9 other species as Over…shed. Two additional species stocks

are classi…ed as Approaching Over…shed Condition. Controlled access regulations reduce

…shing pressure but raise ‡eet harvesting costs. Periodic …shery closures create a race for

…sh that has led to marketing gluts and depressed dockside prices. The race for …sh also

lures vessels to sea in hazardous weather conditions and has been cited as the cause of

recent vessels sinkings.1 Managers and industry participants agree that the commercial reef

…sh …shery is in desperate need of management reform. However, a consensus has not been

reached as to which management alternative is best.

This paper provides an ex ante estimate of potential economic bene…ts under tradable

harvest permits, one of the management alternatives that is being considered by indus-

try and the Gulf of Mexico Fisheries Management Council.2 Under tradable permits, or

property rights-based (RB) management, the aggregate catch is controlled by allocating

exclusive rights to harvest speci…ed quantities of reef …sh during each harvest season. It

is well known that RB programs provide incentives to exploit common pool resources ef-

…ciently, that is, to extract the largest resource rent from harvesting the target aggregate

catch (Montgomery, 1972). However, managers and industry contemplating RB programs

must do so without knowing the extent of the economic bene…ts that are available. An ex

1On April 2, 2001, the vessel Wayne’s Pain sank in bad weather, 85 miles o¤ Marsh Island, LA. Thevessel’s captain stated, “I wouldn’t have been out there then except the derby was on” (Washington Bureaureport, 2001). The derby referes to a 10 day season opening for red snapper which began on April 1.

2A recent study by Dupont (2000) estimates the potential gains from adopting individual transferablevessel quotas in the paci…c salmon …shery. Other empirical studies that predict the gains from propertyrights based …sheries management programs include Squires, Alauddin and Kirkley (1994), and Squires andKirkley (1991).

2

ante estimate of the potential gains thus provides vital information that will undoubtedly

assist in the management reform process.

To estimate the value of the reef …sh …shery prior to implementation of RB management,

we simulate the harvesting and marketing practices that are expected to prevail under a

system of well-de…ned harvest rights for reef …sh. Experience with other …sheries that

have adopted RB programs, discussions with industry and managers, a survey of reef …sh

…shers, and careful attention to the economic incentives that are implicit under a system

of tradable harvest permits are used to accomplish this …rst step. Conditional on the

predicted harvesting and marketing practices, we estimate the net revenues from harvesting

the aggregate reef …sh catch in 1993, the year of our data. The projected net revenues are

compared to the realized net revenues under the controlled access program to obtain an

estimate of the potential bene…ts of management reform.3

The analysis reveals two main sources of economic gain. First, on the revenue side,

elimination of the seasonal closure policy, currently used to conserve red snapper stocks, will

allow …shers to spread the aggregate catch more evenly throughout each year. For example,

the red snapper …shery was open for 95 of a possible 365 days in 1993. This caused large

quantities of red snapper to reach consumer markets in a short time period which reduced

the average dockside price. Eliminating red snapper market gluts is estimated to generate

an annual revenue gain between $2.245m and $3.890m in 1993 (all values are reported in

1993 dollars).

A second major source of economic gain is harvest cost savings that will result as per-trip

catch limits and seasonal closures are replaced with secure harvest permits. The opportunity

cost of owning tradable harvest permits provides incentives to exploit all available economies

of scale and scope in production. Reef …sh harvesting will be carried out by cost e¢cient

vessels operating at e¢cient scale and scope resulting in a signi…cant reduction in the ‡eet

size. In 1993 for example, 387 vessels harvested reef …sh in the northern Gulf region.

3 It is possible that new marketing and harvesting practices will emerge when the …shery is switched to RBmanagement. Such new practices will be adopted only if they are more pro…table than currently availableones. Our estimates thus provide a lower bound on the rent gains available under RB management.

3

We predict that between 31-73 vessels are capable of harvesting the entire 1993 reef …sh

catch. The accompanying harvest cost savings are estimated to range between $8.176m and

$8.758m in 1993.

The combined 1993 revenue gains and harvest cost savings imply economic bene…ts

in the commercial sector of $11.614m with a range of $10.421m and $12.648m.4 Bene…ts

similar in magnitude were likely available throughout the controlled access management

regime, 1990-2001. These numbers suggest that RB management should be given serious

consideration for reef …sh in the northern Gulf.

The remainder of the paper is organized as follows. The next section summarizes the

regulatory history in the northern Gulf commercial reef …sh …shery. A strategy to predict

vessel harvesting and marketing activities under RB management is outlined. Section 3

presents a directional distance function (DDF) model of a multiproduct harvest technol-

ogy. The DDF is used to study harvesting e¢ciency under controlled access management

and to predict harvest activities under tradable harvest permits. Data, estimation, results

and discussion appear in Section 4. Section 5 summarizes the main …ndings and provides

concluding remarks.

2 Industry background and regulations

The reef …sh …shery in the Gulf of Mexico is a complex of bottom-dwelling species consisting

of red, vermilion, and other snapper species, yellowedge, gag, warsaw and other species of

groupers, amberjacks, trigger…sh, porgies, tile…sh, and a host of others. The component

of the reef …sh …shery studied in this paper extends through the northern region of the

Gulf of Mexico, west of Panama City, Florida to the Texas-Mexico border. We note that

the management problems in the eastern Gulf commercial reef …sh …shery are similar in

character to those in the west. Space limitations do not permit a comprehensive analysis of

the bene…ts of management reform in the eastern region, and is reserved for future work.

4We should note that these bene…ts are available only if managers and industry can design and implementa functioning rights-based management program.

4

An average of 467 vessels participated in the northern reef …sh …shery during 1990-2001.

A typical commercial reef …shing trip in the northern Gulf involves steaming to a selected

…shing site which may be located in excess of 100 miles from port of departure. A weighted

vertical line containing several baited hooks is lowered to the desired depth. Once reef …sh

are lured to the bait the line is recovered either with a hydraulic or hand-powered winch.

Vessels often search multiple sites on a single trip. When the vessel captain is satis…ed

with the harvest quantity, or possibly when the quantity harvested reaches the per-trip

catch limit under controlled access management, the vessel steams back to port and sells

the catch. In the absence of per-trip catch limits, most trips average less than 7 days with

median trip length equal to 3-4 days. The main (primary) inputs used to harvest reef …sh

are the capital services provided by the vessel, fuel used during steaming and …shing, labor

services provided by the captain and crew, bait and ice, and food and supplies to sustain

the captain and crew during the trip.

The highest-valued reef …sh species in the northern Gulf is red snapper, which has also

been the most regulated species. A Gulf of Mexico Reef Fish Fishery Management Plan,

hereafter FMP, was implemented in November 1984. The FMP enacted a few simple har-

vesting regulations that prohibited …shing practices considered to be destructive to the

marine environment. Minimum size restrictions (12 inches fork length) were adopted for

commercial and recreational red snapper. Since enactment of the original FMP, a series of

amendments have been adopted, and reporting requirements were introduced to assist ongo-

ing management. The management actions contained in these amendments vary but most

take additional measures to reduce …shing pressure by an increasingly skilled commercial

‡eet and an expanding for-hire and recreational sector.

Amendment 1 set a total allowable catch (TAC) for red snapper allocating 51% to

the commercial sector and 49% to the recreational sector. The commercial TAC was set

at 3.10m pounds in 1990, was reduced to 2.04m pounds from 1991-92, increased to 3.06m

pounds during 1993-95 and increased again to 4.65m pounds during 1996-2000. Amendment

4 (May, 1992) imposed a temporary moratorium on new commercial reef …sh entry permits.

This moratorium has been extended several times. The intent of Amendment 4 was to

5

moderate increases in …shing e¤ort and to stabilize …shing mortality while the Council

considered a more comprehensive e¤ort limitation program. An emergency rule, e¤ective

December 30, 1992, created a red snapper endorsement system that restricted per-trip vessel

catch quantities. Reef …sh permit holders who could demonstrate a harvest quantity of at

least 5,000 pounds of red snapper in two of the three years during 1990-92 were granted

an endorsement which entitled the vessel to catch a maximum of 2,000 pounds of red

snapper per trip. All other quali…ed vessels were restricted to harvest a maximum of 200

pounds of red snapper per trip. The endorsement system was in place from 1993-1997. In

1998, a licensing system was adopted, which granted 138 class 1 licenses to endorsement

permit holders and 8 other qualifying …shers. Class 2 permits were granted to 559 vessel

owner/operators. Class 1 licenses allow 2,000 pounds of red snapper per trip, while class 2

licenses allow a maximum of 200 pounds of red snapper per trip.

Fishery closures were used with the endorsement system to ensure that the commercial

red snapper TAC was not exceeded. Closures initiated a derby …shery which is considered

responsible for many of the current management problems (Thomas, et. al 1993; Waters,

2001). In 1990, the red snapper season was open for 365 days. From 1991 through 2000,

the red snapper season has remained open for 236, 95, 95, 78, 51, 77, 74, 72, 70 and 76

days, respectively.

The Council proposed an individual transferable quota management program for the

commercial red snapper …shery, to begin in 1996. However, a Congressional action in late

1995 prohibited its implementation. Congress imposed a moratorium on RB management

in US …sheries pending a thorough review of their e¤ects. This moratorium is extended

through 2002.

2.1 Economic distortions under controlled access management

The race for …sh and periodic seasons closures cause large quantities of red snapper to reach

consumer markets in a short period of time, resulting in market gluts and reduced dockside

prices (see also Casey et. al, 1993; Geen and Nayer, 1988; Gauvin et. al, 1994). From

6

1986-89 the annual average red snapper prices remained stable at roughly $2.75 per pound.

During a four year period from 1991-94, annual average red snapper prices averaged only

$2.10 per pound.

Red snapper regulations also a¤ect …shing practices. Reef …sh …shermen were asked,

with an open-ended question, how federal …shing regulations have changed their …shing

activities (see Waters, 1996 for details). Nine of the 49 respondents (18.3%) indicate they

take shorter and more trips due to per-trip catch limits. Because a large component of

harvest costs is incurred steaming from port to the …shing ground and back, shortened

trips increase costs per unit of harvest. Industry sources indicate that medium sized vessels

(30-40 feet in length) are capable of transporting up to 5,000 pounds of properly iced …sh.

Smaller vessels can transport between 2,000-4,000 pounds, while large vessels can properly

ice up to 50,000 pounds of …sh.5 Thus, the per-trip catch limit restricts red snapper harvest

for many vessels operating in the …shery.

When the red snapper season is closed, vessel services either remain idle or are allocated

to harvest other reef …sh species. The number of trips that a vessel can take during a …xed

calendar period is to some extent constrained by the time required for steaming to the

grounds, vessel o¤-loading and replenishment of supplies, vessel maintenance, and a rest

period for the captain and crew. While total number of trips per year varies due to weather

and possibly unforeseen maintenance delays, the number of trips that a vessel can take per

year typically ranges between 40-45. With an average trip length of 3-4 days plus a 3-4 day

rest period, vessels are capable of …shing 120-135 days per year. Seasonal lengths in the

range of 50-100 days thus restrict the allocation of …shing e¤ort to red snapper production.

Discussions with industry suggest that the derby environment under seasonal closures

encourage some vessels to increase crew size, although this is a more recent phenomenon.

An additional crew member can speed harvesting operations and eliminate the need to stop

…shing while the crew rests; an additional crew member enables shift work where one crew

member sleeps while others continue …shing operations. This practice must be assumed

5Felix Cox, personal communication 2001.

7

privately optimal under the closure policy but input allocatively ine¢cient in the absence

of harvest time constraints.

Under the seasonal closure policy, …shermen holding red snapper endorsements regularly

target red snapper exclusively during openings and targeted other species exclusively during

red snapper closures. Indications from industry participants and the cost survey data

suggest that some vessels incur nontrivial switching costs to harvest dissimilar species. For

instance, the cost required to switch from red snapper to king mackerel, a demersal species,

are reported to be $585 on average.

The main sources of economic gains under tradable harvest permit management are

expected to be increased revenues from eliminating market gluts, and reduced harvesting

costs as reef …sh vessels exploit economies of scale and scope in production. To examine

these harvest cost savings systematically, the next section introduces a directional distance

function model of the harvest technology. The model is used to characterize the harvesting

ine¢ciency under controlled access management, and to measure the cost savings under

tradable harvest permits. An estimate of the anticipated revenue gains follows in Section 4.

3 Model of the harvest technology

Consider a representative vessel that allocates inputs x 2 <N+ to produce outputs y 2 <M+during a given calendar period. The production set de…nes feasible input and output com-

binations, (x; y), which will be referred to as the vessel activity level. In a regulated …shery,

feasible input-output combinations will depend on stock abundance and the regulations

used to prevent over…shing. Denote the feasible set as

T (S;R) = f(x; y)j x can produce y given S and Rg;(1)

where S is an index of stock abundance and R denotes the regulations. The feasible set is

assumed to be closed and convex. Input and output disposibility assumptions are discussed

below.

8

The directional distance function (DDF) is a functional representation of the feasible

set (see Chambers, Chung and Färe, 1996),

¡!D (x; y; gx; gy) = maxf¯ 2 <j(x ¡ ¯gx; y + ¯gy) 2 T (S;R)g;(2)

where gx 2 <N+ , gy 2 <M+ , g ´ (gx; gy) 6= (0N , 0M), is a directional vector. The DDF

gives the maximal translation of the activity (x; y) in the reference direction (¡gx;gy) that

keeps the translated activity in the feasible set. It should be emphasized that the DDF in

equation (2) is conditional on stock abundance and regulation since it is de…ned relative to

the set T (S;R). This dependance is suppressed only for notational convenience.

The DDF is a complete functional representation of the technology and provides a

convenient measure of technical e¢ciency. When ¡!D (x; yjgx; gy) = 0, no feasible translation

of (x; y) is possible. When ¡!D (x; yjgx; gy) > 0 activity (x; y) is located in the interior of

T (S;R). The technically e¢cient activity level is obtained directly as (x¡¡!D (x; yjgx; gy)gx;y + ¡!D (x; yjgx; gy)gy). Commonly used radial distance functions are special cases of the

DDF. In particular, the Shephard input distance function, Di(x; y), is a special case of the

DDF.6 At directional vectors gx = x and gy = 0M , it can shown that ¡!D (x; yjx; 0M) =

1 ¡ Di(x; y)¡1 (see Chambers, Chung, and Färe, 1996, and Färe and Grosskopf, 2000 for

additional discussion).

The minimum cost function is expressed as

C(y;w) = minx

nw

³x ¡ ¡!D (x; yjx; 0M) ¢ x

´o:(3)

A measure of cost e¢ciency is obtained as the ratio of the frontier and actual cost incurred by

a …shing vessel; O(y; x; w) = C(y;w)=wx. If O(y; x; w) takes the value 1, then activity (x; y)

is located on the cost frontier. The cost e¢ciency measure O(y; x; w) can be decomposed

6The Shephard input distance function is de…ned as Di(x; y) = supf¸ : (x=¸; y) 2 Tg.

9

as (see Färe, Grosskopf and Lovell, 1994),

O(y; x; w) = Di(x; y)¡1 ¢ AE(x; y;w);(4)

where AE(y; x; w) is the component of cost ine¢ciency that results from employing the

wrong input mix given input prices w. When Di(x; y)¡1 = 1, activity (x; y) is technically

e¢cient in that a radial contraction of x is not an element of the feasible production set.

When AE(x; y; w) = 1, x is input-allocatively e¢cient given output y and input prices

w. Clearly, only if Di(x; y)¡1 = AE(x; y; w) = 1 is activity (x; y) located on the cost

frontier. Techniques for analyzing and decomposing scale e¢ciency (or scale ine¢ciency)

are available (see Färe, Grosskopf and Lovell, 1994). Implementation of these measures in

our empirical data is discussed in the following section.

4 Data and empirical estimation

Our data are from an extensive cost survey that was conducted for the 1993 harvest season

(see Waters, 1996 for a detailed description of the cost survey design and data). The cost

survey elicited information through personal interviews with 196 vessel operators, of which

99 operated in the northern Gulf region. Cost survey data are linked to the National Marine

Fisheries Service Log Book reporting system which maintains detailed records of trip-level

quantities harvested, trip lengths and location, and the number of crew members on board

the vessel.7

Data indicate that the 99 sample vessels harvested a total of 2.69m pounds of …sh in

1993, consisting of 90 di¤erent species. While the number of species is large, a much smaller

group of species make up the bulk of the total catch. Red snapper represents the largest

component accounting for 47.4% of the total harvest. The 10 largest volume species account

for 88.7% of the sample vessel catch.

7Fishing activities and costs may di¤er by region if …shers target regional stock concentrations. The datawere initially subdivided by the county in which …shing trips originate. A comparative analysis indicatedno signi…cant county-level di¤erences in production activities.

10

Per-trip data (1443 obs.) Seasonal data (99 obs.)Mean Std. dev. Max. Mean Std. dev. Max.

Red snapper (y1) 883.97 975.69 6,314.35 12,884.55 19,432.53 75,458.91Other reef …sh (y2) 713.13 1673.26 21,161.53 10,394.46 16,883.93 87,025.05Non-reef …sh (y3) 320.53 1,067.43 8,762.50 4,671.99 11,248.37 60,056.47

All …sh 1,917.64 1,777.53 21,161.53 27,951.00 34,043.48 123,166.91

Mean Std. dev. Max. Mean Std. dev. Max.Trips - - - 14.57 13.28 47

Days at Sea 3.79 2.82 14 44.19 43.38 170Vessel services (x1) 178.18 163.05 910.00 2,157.17 2,417.96 12,750.00

Labor (x2) 14.06 13.72 66.5 167.32 194.59 1,020.00Fuel (x3) 243.07 212.76 1200.00 3,358.62 4,188.92 24,000.00

Table 1: Descriptive Statistics for 1993 Reef Fish Vessels.

The harvest species are aggregated into three output groups. To minimize bias caused

by aggregation, industry participants and managers were consulted to identify species that

are harvested using similar methods, thus requiring similar inputs. It was determined that

species within the reef …sh complex can be found at di¤erent depths and may be more active

at di¤erent times of day, but are harvested at similar locations using the same hook and line

technology. In contrast, non-reef …sh species such as king mackerel, yellow…n tuna, bonito,

and wahoo are harvested at di¤erent locations using di¤erent methods.

The …rst output category, y1, is red snapper. In addition to comprising the largest

percentage of total catch, red snapper is subject to regulations not imposed for other reef

…sh species. The second output category, y2, aggregates 80 other reef …sh species. Within

this group, vermilion snapper (28.9%), yellowedge grouper (19.8%), and greater amberjack

(6.4%), make up the largest component. The third output category, y3, includes all non-reef

…sh, which is comprised primarily of king mackerel (69.7%), and yellow…n tuna (24.1%).

Table 1 reports average output and standard deviation of output for the 99 sample

vessels. Recall that the 1993 data were generated under red snapper harvest restrictions.

Nonetheless, on average, vessels harvested more red snapper per trip and per season than

all other reef …sh species. Table 1 does reveal evidence of the e¤ects of the regulatory

environment. The maximum red snapper harvest quantity per trip is small compared to

11

the maximum harvest per trip of other reef …sh species.8 Evidence of the e¤ects of season

closures is also indicated. The maximum seasonal quantity of red snapper is less than the

maximum seasonal quantity of other reef …sh.

Heterogeneity in vessel activity levels is a conspicuous feature of the data. The output

mix at the trip level and at the seasonal level varies widely across vessels. Some of this

variation can be attributed to di¤erences in species mix across regions. However, large

variation in species mix is also consistent with species-speci…c regulations, such as ownership

of 2,000 versus 200 pound per trip red snapper permits.

The data also exhibit considerable variation in the per vessel and seasonal total catch

quantities. The maximum total catch per-trip for all species is 11 times larger than the per

trip average, and the maximum total catch per season is 4.4 times larger than the seasonal

average. Large variation in the number of trips per season further illustrates large variation

in the vessel scale of operation.9

Table 1 reports the average and standard deviation for the number of trips, days at sea,

and the three main inputs used by reef …sh vessels. We assume that ice and bait use is

proportional to the quantity harvested and that food for the crew is proportional to the

labor input. The services that are provided by the vessel capital, x1, are measured as the

vessel length in feet times the number of days that the vessel spent at sea. The fuel input,

x2, is measured in gallons, and labor x3 is measured as the number of crew on board the

vessel times the days the vessel is at sea.10

The price of vessel capital services and fuel are obtained from the cost survey data. The

unit price of labor services is di¢cult to assess in the reef …sh …shery because crew are

paid using a revenue share system, and because skilled crew members typically earn a wage

8Nine of the 1443 observations indicated red snapper catches that exceed the 2,000 pound limit. Logbook data could not used to penalize operators for exceeding catch limits, and thus violations occurred,although infrequently.

9The cost survey data provides an explanation for this heterogeneity; 19 of the 99 vessels in the sampleearn part of their annual income in non-…shing activities.

10 Input use is not proportional to days at sea. Trip length varies and fuel consumption is heaviest duringthe steam from port to the …shing ground, so that fuel consumption is not proportional to days at sea.Furthermore, our data indicate that 30 vessels adjusted crew sizes during the 1993 …shing season.

12

premium. The cost survey data could not be used to obtain an accurate measure of the

labor remuneration actually paid by vessel operators. We approximate the labor price using

average hourly wage for production workers in the northern Gulf region (US Department

of Labor, 1994). This yielded a crew-member wage per day of $93.96.

For estimation purposes, vessels with incomplete data and vessels that reported less than

three trips during 1993 were dropped. The data were examined to identify and remove

observations with evidence of reporting error.11 There were 71 observations used in the

estimation. Of these, 40 vessels held class 1 red snapper endorsement permits and 31

vessels held class 2 permits. Hereafter, vessels with 2,000 pound red snapper permits are

referred to as class 1 vessels and all other vessels are referred to as class 2 vessels.

An empirical estimate of the DDF is obtained as the solution to the following linear

programming problem

¡!D (x; yjgx; gy) = max¯ 2 <(5)

s.t.Pk zkykm = ym + ¯gy;m; m = 1; 2

Pk zkykm ¸ ym + ¯gy;m; m = 3

Pk zkxkn · xn ¡ ¯gx;n; n = 1; 2; 3Pk zk ¸ 1;

zk ¸ 0; k = 1; :::;K;

where m and n index outputs and inputs, respectively, zk is the intensity variable for vessel

k, and K is the number of vessels in the sample.

The constraints in equation (5) impose two important assumptions for the harvest tech-

nology. First, the harvest technology is assumed to exhibit weak output disposibility for red

snapper and other reef …sh, outputs y1 and y2, respectively. This assumption is imposed

through the constraintPk zkykm = ym + ¯gy;m for m = 1; 2. Recall that reef …sh species

11One vessel reported a per trip red snapper harvest more than three times the 2,000 pound limit (seeTable 1). This vessel observation was dropped.

13

co-habitate and non-targeted species are typically intercepted. This implies that reductions

in the quantity of red snapper, y1, will be accompanied by reductions in the quantity of

other reef …sh, y2, and visa versa. Because non-reef …sh species can be harvested without

intercepting reef …sh, the assumption of strong output disposibility is maintained for the

output y3.

The second assumption is that the harvest technology exhibits non-decreasing returns

to scale. The assumption is imposed by the constraintPk zk ¸ 1. We know that reef …sh

vessels must steam considerable distances from port before …shing begins. Strictly positive

quantities of fuel, vessel capital services and labor are required to harvest even a small

quantity of reef …sh and thus the technology cannot exhibit constant returns to scale. To

understand the assumption of non-decreasing returns to scale, it is essential to recognize

the role of the controlled access management program on the data generating process. The

convex hull of the observed data characterizes the feasible production set under controlled

access management. Per-trip red snapper harvest limits are known to constrain the harvest

of some reef …sh vessels. Switching from controlled access to RB management will likely

expand the feasible production set and T (S;RB) ¶ T (S;CA), where CA is “controlled

access” management. Consequently, basing our analysis on the convex hull of the 1993

data will underestimate production possibilities that are available under RB management.

Figure 1 illustrates.

Figure 1 depicts hypothetical data for four vessels operating under a per-trip catch limit.

For simplicity, we assume a single-input and a single-output. Vessel a employs input xa

and harvests output ya and similarly for vessels b; :::; d. Under constant returns to scale,

an estimate of the feasible set is obtained as the canonical hull of the observed data. In

Figure 1, the boundary of the feasible set is a ray extending from the origin through (xb; yb),

vessel b’s observed activity. Under constant returns to scale and strong input disposibility

activities on and to the right of the segment xaab are feasible.

Under the assumption of variable returns to scale, the boundary of the feasible set is

xaabde and vessels a, b, and d are considered technically e¢cient. Activity d exhibits a low

output/input ratio relative to b, which is consistent with a region of the technology that

14

Figure 1: E¤ects of Per-trip Catch Limits on Data.

exhibits decreasing returns to scale. Importantly however, the low output/input ratio at d

is also consistent with a per-trip catch limit policy, which is more likely to bind for large

vessels. It is important that the relatively low output/input ratio at d not be interpreted

as a region of decreasing returns to scale unless there is reason to believe that the harvest

technology truly exhibits this property.

Discussions with industry indicate that some vessels were not severely constrained by

the per-trip catch limit. For example, suppose vessel b holds a 2,000 pound per trip red

snapper permit and typically harvests less than 2,000 pounds per trip. In this case, activity

15

(xb; yb) is feasible and technically e¢cient with and without the catch limit regulation.

Furthermore, activity (xb; yb) should be replicatable simply by taking additional trips from

port. This implies (2xb; 2yb), point f in Figure 1 is also feasible and technically e¢cient

with and without per trip catch limits.

Based on prior knowledge of the data generating process, and on the assumption that

some vessels in the 1993 data are unconstrained under per-trip catch limits, we assume

that the harvest technology exhibits non-decreasing returns to scale. The frontier that is

attainable under RB management can be approximated by the intersection of the convex

and canonical hulls of the 1993 data. In Figure 1, the feasible set under non-decreasing

returns to scale is de…ned by the segment xaab and then along a ray that extends from b

through f .

Two additional factors should be considered. First, the fact that trips are discrete implies

that a linear combination of (xb; yb) and (2xb; 2yb) is not feasible. Bias introduced by the

discrete nature of the input-output bundles is controlled by eliminating vessel observations

that are comprised of a small number of trips (less that three trips in our case).

Second, the slope of the ray bf may underestimate the feasible output/input ratio for

unconstrained vessels, particularly larger ones. The output per unit of input is zero during

the steam from port to a chosen …shing site. Thereafter, the quantity harvested per unit of

input is positive. It is possible that in the absence of a per-trip catch limit vessel c would have

obtained the activity level (xc0 ; yc0) at point c0. That is, once at the …shing site, vessel c could

have allocated additional inputs xc0 ¡ xc and obtained additional output yc0 ¡ yc. Because

(xc0 ; yc0) is unobserved, it is possible that the RB-regime frontier is underestimated.12 We

are unaware of a means to determine the e¤ect that unobserved activity c0 has on our results,

but note that if anything the results that follow will underestimate the attainable frontier

under RB management.

12We would like to thank Jay Coggins for this insight.

16

1 2 3 4 5¡!D (x; yj0n; gy)gy = (y1; 0; 0)

¡!D (x; yj0n; gy)gy = (0; y2; 0)

¡!D (x; yjx; 0m) O(y; x; w) AE(x; y; w)

All vessels(71 obs.)

0:68(0:45)

0:56(0:44)

0:31(0:32)

0:65(0:33)

0:89(0:15)

Class 1 vessels(40 obs.)

0:84(0:34)

0:48(0:43)

0:18(0:23)

0:73(0:26)

0:89(0:16)

Class 2 vessels(31 obs.)

0:48(0:49)

0:68(0:43)

0:47(0:35)

0:54(0:39)

0:89(0:14)

Table 2: E¢ciency Analysis Results. Class 1 vessels are those in possession of2,000 pound per trip red snapper permit, Class 2 vessels refer to all othervessels. Standard deviations are in parentheses.

4.1 Results

4.1.1 Harvesting e¢ciency under controlled access management

Table 2 reports the sample average and standard deviation for …ve measures of harvesting

e¢ciency for the 1993 sample vessels. For comparison, averages and standard deviations

are reported for subsamples consisting of class 1 and class 2 vessels.

Columns 1 and 2 of Table 2 assess output technical e¢ciency. Column 1 reports the

DDF with directional vectors gx = 0n and gy = (y1; 0; 0). Here the DDF gives the maximum

feasible translation (expansion) of observed red snapper harvest, y1, holding inputs, and

outputs y2, and y3 …xed. The full sample average value is 0.68 with standard deviation

0.45. The interpretation is that the frontier quantity of red snapper is on average 0.68 times

larger than the quantity that was actually harvested. A comparison across vessel classes

indicates that, on average, the frontier quantity for class 2 vessels is 0.48 times larger than

observed while the frontier quantity for class 1 vessels is 0.84 times larger than observed.

Note that ten class 2 vessels did not harvest red snapper (all class 1 vessels harvested red

snapper) and thus obtain a DDF value of zero when gy = (y1; 0; 0). The average value of

the DDF, with gy = (y1; 0; 0), for the 21 class 2 vessels that harvested positive amounts of

red snapper in 1993 is 0.71 with standard deviation 0.44.

17

Column 2 of Table 2 reports the sample average DDF estimate for gx = 0n and gy =

(0; y2; 0). In this case, the DDF measures the maximal translation of the observed other

reef …sh catch, y2, to the frontier holding inputs, and outputs y1, and y3 …xed (all but 2

vessels harvested positive quantities of y2). The full sample average is 0.56 with standard

deviation 0.44. A comparison with the results in column 1 suggests that on average the 1993

‡eet attained higher technical e¢ciency in the harvest of other reef …sh species than red

snapper. This result is consistent with the controlled access regulatory environment where

no signi…cant harvest regulations were directed at the reef …sh species comprising the output

group y2. Notice also that class 1 vessels attained a higher level of technical e¢ciency in

the production of y2 than class 2 vessels. This suggests further that the relatively low level

of technical e¢ciency indicated for red snapper harvesting likely re‡ects the e¤ects of the

per-trip catch limitation.

Column 3 of Table 2 reports the average DDF value for directional vector gx = x and

gy = 0m and thus measures the maximal radial translation of the input bundle to the

frontier holding observed output constant. Column 4 reports the average ratio of frontier

and observed costs, O(x; y; w). The values for both input-based measures indicate that

actual input usage was substantially larger than the frontier quantity. A comparison across

vessel classes …nds that, on average, the input bundle used by class 1 vessels could have been

contracted radially by a factor of 0.18, and the input bundle used by class 2 vessels could

have been contracted radially by a factor of 0.47. Class 1 vessels attained a higher level

of input technical and cost e¢ciency than class 2 vessels. These di¤erences are supported

statistically. A Kolmogorov-Smirnov test rejects, at the 99% con…dence level, the null

hypothesis that the empirical distributions of the input-based technical e¢ciency measure,¡!D (x; yjx; 0m), and the cost e¢ciency measure, O(x; y; w), are the same across vessel classes.

Column 5 reports the average input-allocative e¢ciency score. The sample average is

0.89 with standard deviation of 0.15. No di¤erences are indicated across vessel classes.

Input allocative ine¢ciency does not appear to be a signi…cant contributor to harvesting

ine¢ciency.

18

Scale e¢ciency We next characterize the harvesting ine¢ciency that resulted from ves-

sels harvesting at an ine¢cient scale of operation. Our prior information suggests that

the per-trip catch limit policy for red snapper likely caused vessels to operate in a region

of increasing returns. To assess this possibility we estimate the DDF for directional vec-

tor gx = x and gy = 0m, …rst under the assumption of a non-decreasing returns to scale

technology, and then under the assumption that the technology exhibits constant returns

to scale. The constant returns to scale DDF estimate is obtained as the solution to the

linear program in equation (5) with the constraintPk zk ¸ 1 dropped. Scale e¢ciency is

measured as13

¡!SE(x; y) =1 ¡ ¡!DCRS(x; yjx; 0m)

1 ¡ ¡!DNDRS(x; yjx; 0m)

where the DDF in the denominator (numerator) is estimated under the assumption of non-

decreasing (constant) returns to scale. A value of ¡!SE(x; y) < 1 indicates that the activity

level (x; y) is scale ine¢cient due to increasing returns to scale.

The results indicate that overall, 30 of the 71 sample vessels operated in a region of

increasing returns to scale. Of the 40 class 1 vessels, 27 (67.5%) were scale ine¢cient due to

increasing returns, whereas only 3 of the 31 (9.7%) class 2 vessels were scale ine¢cient due

to increasing returns. Further analysis reveals that scale ine¢ciency was most prevalent for

the larger class 1 vessels. The median class 1 vessel length in the sample is 48 feet (the

median class 2 vessel length is 36 feet). There are 19 class 1 vessels less than 48 feet in

length, and of these 9 (47.4%) harvested in a region of increasing returns. There are 21

class 1 vessels that are 48 feet in length or greater, and of these 18 (85.7%) harvested in a

region of increasing returns to scale.

To summarize, the output speci…c analysis …nds that sample vessels harvested 59.5% of

the frontier quantity of red snapper and 69.4% of the frontier quantity of other reef …sh.

Class 1 vessels, which harvested the bulk of the red snapper, harvested 57.2% of the frontier

13This measure is adopted from Färe, Grosskopf and Lovell (1994) who measure scale e¢ciency as theratio of the variable returns and constant returns Shephard input distance function.

19

red snapper and 73.5% of the frontier quantity of other reef …sh. Most class 1 vessels

operated in a region of increasing returns to scale.

The empirical evidence supports the hypothesized e¤ects of the controlled access man-

agement policy. Species-speci…c regulations reduce the e¢ciency and, as desired for the

management object of stock conservation, the potency of the reef …sh ‡eet. Technical and

scale ine¢ciency is disproportionately concentrated in red snapper harvesting activities de-

spite the fact that the initial allocation of class 1 endorsement permits went to vessels that

could demonstrate an historical record of high catch rates, i.e., class 1 vessels are more likely

to be experienced and successful operators.

4.1.2 Harvest costs savings under RB management

The scenario considered is one where the reef …sh …shery is managed under a system of

tradable harvest permits for all reef …sh species,14 and the harvest ‡eet is in long run

equilibrium. If all vessels harvest the same mix of red snapper and other reef …sh, necessary

conditions for equilibrium are: (1) the marginal harvesting cost be equal for each output

and across all active vessels; and (2) that no active vessel can pro…tably exit or enter the

…shery, in other words, the value of active vessels including the value of the tradable harvest

permit is higher in the …shery than out. The …rst condition ensures that there are no gains

from redistributing the catch among active vessels (see Montgomery, 1972). The second

condition states that the ‡eet is in long run equilibrium.

We next identify the vessel activity level that is expected under RB management. Con-

sider …rst the output mix. Industry sources and information from the cost survey suggest

that there are diseconomies of scope in the production of reef …sh and non-reef …sh species.

Vessels must modify the gear used to harvest non-reef …sh. These modi…cations represents

14Adopting RB management for a single species e.g., red snapper or a subset of species in the complex willcreate other serious management problems. First and foremost, …shing e¤ort that is released under tradableharvest permits will be redirected to reef …sh species that remain under controlled access management.Because …shers cannot target individual species within the reef …sh complex exclusively, bycatch problemswill intensify. And, increased …shing pressure on species that remain under controlled access managementinvites similar problems experienced in red snapper.

20

a superadditive …xed cost that increases total harvest costs. No evidence is found to suggest

cost complementarity between reef …sh and non reef …sh species, and thus we conclude that

diseconomies of scope exist in the production of non-reef …sh (Baumol, Panzar and Willig,

1982).

Cost complementarity between red snapper and other reef …sh species is a natural con-

sequence given the nature of reef …sh …shing. Species within the complex co-habitate and

are typically intercepted by the same gear on the same trip. Consequently, single-species

harvesting will involve discarding non-targeted species. A vessel that attempts to harvest a

single species by discarding a non-targeted …sh must incur additional costs to harvest the

discarded …sh at a later date. This will cause the sum of species-speci…c costs to exceed the

cost of harvesting the same quantity of reef …sh jointly.15

Gear modi…cation costs, in order to harvest non reef …sh species, and additional costs

incurred under specialized reef …sh harvesting will raise average costs and reduce the value

of tradable reef …sh harvest permits. We conclude that the output bundle under tradable

permit management will be comprised of a mix of red snapper and other reef …sh species,

and zero quantity of non-reef …sh species. Predicting the precise mix of red snapper and

other reef …sh is a complex matter that will not be attempted. For convenience we set the

individual vessel output mix equal to the aggregate output mix harvested in 1993, and note

that additional gains from outputs adjustments may be available under tradable harvest

permits.16

Consider next the scale of production for a vessel operating over a full harvest season.

Vessel capital services, i.e., vessel length times days at sea, are constrained by the number of

days in a year and weather conditions. Discussions with industry and evidence from the log

book data indicate that a vessel that is 35 feet in length is capable of spending roughly 100

days at sea in a typical year. Larger vessels can …sh in more severe weather conditions. A

15Equilibrium ‡eet structure under tradable harvest permits and joint production is characterized inSquires and Kirkley (1996) and Weninger (1998).

16Tradable harvest permits under multiple output technologies are considered in Squires and Kirkley(1996) and Weninger (1998).

21

vessel of length 45, 55 and 65 feet can spend roughly 110, 120, 130, days at sea, respectively

in a typical year.

Reef …sh vessel owners incur …xed annual operating costs to secure the services of the

vessel capital. These costs include state and federal license fees, docking fees, o¢ce sta¢ng

and equipment costs. In addition, vessels must occasionally undertake costly hull and engine

overhauls, and replace damaged or worn equipment. Under non-decreasing returns to scale,

the presence of …xed annual costs imply that ray average cost (RAC) will have a traditional

U-shape. RAC will attain a minimum when the input bundle is allocatively e¢cient and

when all available vessel capital services are utilized, i.e., when the vessel spends the maximal

days at sea. Because vessel capital services are constrained for a vessel of …xed size, further

increases in output require increases in the quantity of fuel and/or labor per-unit of vessel

capital. This implies input allocative ine¢ciency. It can be shown that the output that

minimizes global RAC maximizes the value of a vessel operation, and thus maximizes the

residual return to the tradable harvest permits.17 Lastly, tradable harvest permits will be

owned by cost e¢cient vessels.

Vessel activity under RB management The analysis will identify the RB harvest

activity for four representative vessel sizes, indexed j = 1; :::; 4. Vessel 1 is 35 feet in length,

vessels 2, 3, and 4 are respectively, 45, 55, and 65 feet in length.18 The following four step

procedure is used to identify the RB-regime vessel activity levels.

Step 1. The objective of the …rst step is to identify the allocatively e¢cient input mix

required to harvest an output vector that will be denoted yj;1 for vessel j.19 Output yj;1 is

comprised of positive quantities of red snapper and other reef …sh but no pelagic species,

17Let y¤ denote the annual output that minimizes global RAC. Consider an increase in output from y¤ toÀy¤, where À > 1. Revenues increase by a factor of À. But if y¤ minimizes RAC, C(Ày¤) > ÀC(y¤) whichimplies pÀy¤ ¡ C(Ày¤) < py¤ ¡C(y¤). A similar argument can be made for the case where À < 1.

18There are only 7 vessels in the sample less than 35 feet in length, and reliable estimates of …xed costscould not be obtained. The data contained 19 vessels 30-40 feet in length, 15 vessels 40-50 feet in length, 12vessels 50-60 feet in length and 12 vessels greater than 60 feet in length.

19The output yj;1 is selected to ensure that it lies in the interior of the feasible production set. The intensityvariables zk for the simulated vessels were checked to ensure that they did not in‡uence the estimate of thefeasible set.

22

i.e., yj3 = 0. The ratio of red snapper and other reef …sh is set equal to the ratio of the TAC

of red snapper and aggregate harvest of other reef …sh in northern Gulf reef …sh …shery in

1993. The allocatively e¢cient input vector is identi…ed using sample average input prices,

and is denoted xj;1.

Step 2. The objective of the second step is to identify the frontier output that can

be harvested using inputs xj;1.20 To accomplish step 2 we estimate ¡!D (xj;1; yj;1j0; yj;1) and

calculate the frontier output yj;2 = yj;1+¡!D (xj;1; yj;1j0; yj;1)yj;1. Denote the technical, input

allocative, and scale e¢cient activity that obtains from step 2 as (xj;2; yj;2).

Step 3. The objective of step 3 is to estimate the harvest costs associated with activity

(xj;2; yj;2). For this purpose it is important to note that harvest cost ine¢ciency will in

general be caused by external factors that are outside of the control of a …rm, such as

distorting regulations and depleted stocks, and internal factors, such as errors in judgement

by the vessel captain and crew (see Freid, Schmidt and Yaisawarng, 1999 for additional

discussion). Regulations that are intended to reduce …shers’ capability to harvest …sh will

be eliminated under RB management. However, ine¢ciency from internal sources will not

be eliminated by changing the operating rules for vessels. Errors in judgement will occur

under the RB management regime and active vessels will operate in the interior of the

feasible production set.

An unbiased estimate of harvest costs under RB management must be purged of ine¢-

ciency caused by distorting regulations but allow for ine¢ciency caused by internal factors.

Furthermore, to allow for the possibility that cost e¢ciency is systematically related to the

output mix, an estimate of internal cost e¢ciency from vessels harvesting both red snapper

and other reef …sh is preferred. To obtain such an estimate we rely on the level of cost

e¢ciency attained by vessels with 2,000 pound per trip red snapper permits. While these

vessels faced a per-trip catch limit they provide a reasonable (best available) estimate of

the internal harvesting e¢ciency that will prevail under RB management.

20Notice that under a constant returns to scale harvest technology, step 2 is unnecessary since (xj;1; yj;1)would already lie on the scale e¢cient frontier. Under a non-decreasing returns to scale technology an outputexpansion will be possible in general.

23

Figure 2: Cost E¢ciency Estimates for Sample Vessels.

Let bORB denote an estimate of internal cost e¢ciency under RB management. Class

1 cost e¢ciency scores are viewed as random sample of identically distributed random

variables that are drawn from a common distribution. A naive bootstrap procedure is

used to estimate percentile values of the cost e¢ciency distribution.21 Figure 2 depicts the

estimated cost e¢ciency scores for all sample vessels and reports the 50’th through 70’th

percentile bootstrap values attained by class 1 vessels.

Simple manipulation of equation (4) obtains the following estimate of harvesting costs

21The naive bootstrap is not appropriate for constructing unbiased con…dence intervals for actual individ-ual vessel e¢ciency scores (Simar and Wilson, 2000).

24

Harvest Quantity Fixed and Average Cost for RB VesselsVessel Length Red

SnapperOther Reef

Fish FC RAC500th Perct. bORB

RAC600th Perct. bORB

RAC700th Perct bORB

35 feet 46,589 96,701 9,017 0.300 0.265 0.24545 feet 65,890 136,763 11,535 0.294 0.259 0.23955 feet 87,853 182,350 14,680 0.292 0.256 0.23665 feet 112,478 233,464 19,478 0.294 0.258 0.238

Table 3: Predicted Vessel Activity, Fixed and Average Cost Under RB Manage-ment FC is annual fixed cost, RAC denotes ray average cost.

under RB management;

CRB(yj;2; w) =Piwix

j;2

bORB; j = 1; :::; 4:(6)

Step 4. The objective of the …nal step 4 is to scale the vessel activity (xj;2; yj;2) and

associated harvest costs CRB(yj;2; w) to the level that is expected in the absence of periodic

seasonal closures. A scaling factor ¸j is calculated such that the quantity of vessel capital

services is equal to the maximal capital services that are available in a harvest season lasting

365 days. For each vessel class the scale factor ¸j is used to obtain the seasonal activity

level ¸j(xj;2; yj;2). Seasonal harvest costs are then obtained as ¸jCRB(yj;2; w).

Table 3 reports the predicted RB-regime output of red snapper and other reef …sh,

annual …xed costs, and RAC for the four representative vessel sizes. The RAC estimates,

the last three columns of Table 3, assume three levels of (internal) cost ine¢ciency with

assumed e¢ciency increasing from left to right. The highest reported RAC assumes that

the RB ‡eet will achieve the 50’th percentile value of cost e¢ciency obtained by the 1993

sample of class 1 vessels; bORB = 0:746. The last two columns in Table 3 assume that the RB

‡eet will achieve, respectively, the 60’th ( bORB = 0:877) and 70’th ( bORB = 0:975) percentile

values.

From Table 3, predicted output increases with vessel size as expected and 55 foot vessels

attain the lowest RAC, although the di¤erences in RAC across vessel sizes is less than one

cent per pound. The estimated RAC of 35 foot vessels is roughly $0.008 per pound higher

25

than the other vessel sizes. The analysis thus predicts that the RB ‡eet will be comprised

of vessels that are between 45-65 feet in length.22

Suppose that the RB ‡eet attains the 60’th percentile level of cost e¢ciency and is

comprised of 45-65 foot vessels. The equilibrium number of vessels is then predicted by

dividing the aggregate catch of reef …sh by the RB-regime output quantities (Table 3).

This calculation indicates that the number of active vessels under tradable harvest permit

management will be 31 if comprised of 65 foot boats and 73 if comprised of 35 foot boats.

Assuming the 60’th percentile level of cost e¢ciency, the RAC for the predicted RB ‡eet

is, on average, $0.258. The total cost of harvesting the TAC of red snapper and aggregate

catch of other reef …sh is estimated to be $2.698m. If the RB ‡eet attains the 50’th (70’th)

percentile level of cost e¢ciency the RAC is predicted to be on average $.293 ($.238) and

the total cost incurred by the RB ‡eet is $3.068m ($2.486m).

Under the controlled access management program, the RAC for the 71 vessels in our

sample is $1.043. Two factors in‡uencing this sample average must be considered. First,

41 sampled vessels harvested non-reef …sh species in 1993. To control for the in‡uence of

non-reef …sh harvest costs we approximate costs associated with reef …sh …shing as the total

cost weighted by the proportion of reef …sh that each vessel harvested. This correction

increases the sample RAC to $1.089. A second consideration is that some vessels in the

1993 sample are part time …shers who take a relatively small number of trips per year. The

sample average RAC for full time reef …sh vessels is $1.030. We correct for non-reef …sh

harvest as above to obtain an estimate of RAC of $1.074.

Using $1.074 as the actual RAC incurred by the 1993 ‡eet, the actual costs to harvest the

aggregate reef …sh catch in 1993 under the controlled access management policy is estimated

to be $11.244m. The predicted costs savings from adopting tradable harvest permits is thus

estimated to be $8.547m if vessels attain the 60’th percentile level of internal cost e¢ciency.

If vessels attain the 50’th (70’th) percentile level of cost e¢ciency, the harvest cost saving

22The …nding that no single vessel size class dominates in terms of RAC is expected. For example, if 45foot vessels attained a lower RAC than other vessels sizes, a reef …sh ‡eet made up of 45 foot vessels wouldbe expected.

26

is estimated to be $8.176m ($8.758m).

4.1.3 Red snapper revenues under RB management

An estimate of the revenue gains under RB management is obtained by predicting the

red snapper dockside price in the absence of seasonal closures. Data on annual average

dockside prices for red snapper were collected from the National Marine Fisheries Service

for 1962-1999. These data are used in the following regression model,23

pt = ®0 + ®1pt¡1 + ®2¿ + ®3DTACt TACt + ®4µt + ²t:(7)

The dependant variable, pt, is the real annual average dockside price for red snapper in year

t. The …rst two right-hand side variables in equation (7) are the lagged red snapper price,

pt¡1, and a time trend, ¿ . The commercial red snapper harvest was managed under a TAC

policy from 1990-1999 and thus aggregate red snapper harvest can be assumed exogenous

during this period; the third explanatory variable is an indicator variable, DTACt = 0 for

t < 1990 and DTACt = 1 for t > 1990, interacted with the TAC. The fourth right-hand

variable µt indicates the proportion of the year that the red snapper …shery remained open

for harvesting. The disturbance term ²t is assumed to have zero mean and …nite variance.

The unknown parameters are estimated using ordinary least squares regression. Parameter

estimates and standard errors are reported in Table 4.

All slope parameters are statistically di¤erent from zero at or above the 90% con…dence

level. Parameter signs are as expected. A larger aggregate harvest leads to lower dockside

prices, (®3 < 0) and increases in the season length raise the average dockside price (®4 > 0).

The adjusted R-square statistic is 0.968. The variance of the error is 0:090. A test of …rst-

order serial correlation is inconclusive (the Durbin-Watson statistic is 1.462).

A price forecast and variance function is calculated for the period 1990-1999 using the

23An alternative approach for estimating the equilibrium red snapper price is to specify and estimate astructural model of red snapper supply and consumer demand. Data limitations, and complications thatarise in modeling supply under a complex regulatory environment precluded such an approach.

27

Variable Parameter Estimate Std. ErrorConstant ®0 -0.156 0.153pt¡1 ®1 0.525¤¤ 0.082¿ (trend) ®2 0.032¤¤ 0.006DTACt TACt ®3 -0.061¤ 0.032µt (season length) ®4 0.709¤¤ 0.154

Table 4: Parameter Estimates for Dockside Price Model There are 37 observations.The adjusted R-square statistic is 0.968. A single asterisk indicates theestimated parameter is significantly different from zero at or above the90 percent level of confidence. A double asterisk indicates statisticalsignificance at or above the 99 percent confidence level.

…tted parameters from Table 4 and by setting µt = 1. The …tted regression line, price

forecast, and 95% con…dence interval for the predicted price are plotted in Figure 3.

The average dockside price for red snapper in 1993 with season length restricted to 95

days was $1.94. The price prediction with season length set at 365 days is $2.84 with a

95% con…dence interval ($2.60, $3.08). The total red snapper catch in 1993 was 3.405m

pounds. Red snapper revenues absent a seasonal closure are estimated to be $9.665m with

95% con…dence interval of $8.843m and $10.488m. The actual 1993 revenue was $6.598m

indicating revenue gains under RB management of $3.068m with a 95% con…dence interval

of $2.245m and $3.890m.

4.2 Discussion

The combined revenue gains and cost savings from harvesting the 1993 reef …sh harvest is

estimated to be $11.614m. There are several factors that can in‡uence this estimate. The

predicted output quantities for vessels under RB management may be less than reported

in Table 3, in which case, additional vessels will be required to harvest the TAC, RAC will

increase, and ‡eet costs will be higher. Suppose that RB-regime vessel spend 10% fewer

days at sea than are assumed above. In this case, the bene…ts of management reform decline

to $11.549m. The di¤erence is small because RAC is relatively ‡at over the range of output

quantities that are indicated by the above analysis.

Our estimate of the cost savings is sensitive to the predicted level of internal cost ef-

28

Figure 3: Annual Average Dockside Prices for Red Snapper in the Gulf of Mexico.

…ciency under RB. An overestimate of RB-regime cost e¢ciency will lead to an in‡ated

estimate of the cost savings due to management reform. Referring to Figure 2 however,

notice that 16 of the 40 class 1 sample vessels (40%) achieved a cost e¢ciency score above

the 60’th percentile value. The sample of 99 vessels that participated in the cost survey

represents roughly one quarter of the 1993 ‡eet, which consisted of 387 vessels. If the cost ef-

…ciency attained by the sample is representative of the 1993 vessel population, then roughly

155 vessels should be capable of achieving the internal cost e¢ciency that is assumed in

our analysis. If anything, our RB-regime ‡eet cost estimate is conservative, and more cost

savings may be realized under RB management.

29

The reported estimate of the bene…ts of RB management reform do not include addi-

tional costs that may be required to administer and monitor the tradable harvest permit

program. RB programs may require additional monitoring of harvesting activities to ensure

harvest quotas are not exceeded (Copes, 1986). Increased monitoring costs detract from

total bene…ts of RB programs. On the other hand, frequent regulatory adjustment is an

unavoidable feature of many controlled access management programs and these adjustments

are not necessary under RB management.24

We have based our estimate of the harvest cost savings on the equilibrium ‡eet structure

that is predicted to emerge under tradable harvest permits. The predicted harvest cost

savings are fully realized only when gains from permit trading are exhausted and ‡eet

downsizing is complete. Delays in the transition to the RB-regime equilibrium could take

several years, and may be sensitive to the method used to initially distribute tradable

harvest permits (see Weninger and Just, 1997).

The above analysis does not address complications that are likely to arise if vessels

which exit the northern Gulf reef …sh …shery under RB management relocate and increase

harvesting pressure on adjacent …sheries. The analysis predicts that more than 300 vessels

will exit the northern Gulf reef …sh …shery under tradable harvest permit management. A

likely …shery for vessel relocation is the eastern Gulf of Mexico reef …sh …shery. The eastern

reef …sh …shery currently faces similar management problems as in the northern region with

controlled access regulations being implemented to rebuild over…shed stocks of key species

such as red grouper and gag. Ensuring that vessels that exit the northern reef …sh …shery

do not exacerbate management problems in other …sheries presents a serious challenge for

managers.

Lastly, we have focused on the 1993 harvest season, the year the cost survey data is avail-

able. Economic distortions under controlled access management have persisted from 1991-

2001. Figure 3 illustrates that dockside red snapper prices have been depressed throughout

24Clay Heaton of the Mid-Atlantic Fisheries Management Council reports that the switch to individualtransferable quota management in the surf clam and ocean quahog …shery led to a signi…cant reduction inthe hours their o¢ce sta¤ dedicated to clam …shery management.

30

this period. The estimated bene…ts for 1993 cannot be extrapolated directly to the other

years that the controlled access management program has been in place. Nonetheless, it is

reasonable to think that bene…ts similar in magnitude to the 1993 estimates were available

throughout the period of controlled access management.

5 Conclusion

This article estimates potential economic bene…ts of switching from controlled access man-

agement to a property rights-based management program in the northern Gulf of Mexico

reef …sh …shery. Our results indicate that spreading the red snapper harvest more evenly

throughout the year in 1993 would have eliminated market gluts, and raised average dock-

side revenue by $3.068m. A 95% con…dence interval of these revenue gains is [$2.245m,

$3.890m]. In addition, signi…cant harvest cost savings could have been generated under

tradable harvest permits. Eliminating per trip catch limits and seasonal closures, and real-

locating harvesting responsibilities to cost e¢cient vessels could have reduced ‡eet harvest

costs in 1993 by roughly 76%. Harvest cost savings of $8.547m could have been realized.

The combined revenue gains and cost savings are estimated at $11.614m, with a range of

[$10.421m, $12.648m]. The northern Gulf reef …sh …shery has been managed under dis-

torting regulations from 1991-2001, and economic bene…ts of similar magnitude were likely

available throughout this period.

Our estimate of the potential rent gains assumes that a system of freely tradable harvest

permits can be designed and implemented. Designing such a program is likely to be com-

plicated, and implementing such a program may be more di¢cult (Committee to Review

Individual Fishing Quotas, 1999; Squires et al. 1998). These obstacles can be weighed

against the economic bene…ts of management reform, which appear to be large.

31

6 References

Baumol, W. J., J. C. Panzar and R. D. Willig. Contestable Markets and the Theory of

Industry Structure. Harcourt Brace, Jovanovich, New York, NY, 1982.

Casey K. E., C. M. Dewees, B. R. Turris and J. E. Wilen. The E¤ects of Individual Vessel

Quotas in the British Columbia Halibut Fishery, Marine Resource Economics, 10(1993):

211-230.

Chambers, Chung and Färe. Bene…t and Distance Functions, Journal of Economic Theory,

70(1996): 407-419.

Committee to Review Individual Fishing Quotas. Sharing the Fish: Toward a National

Policy on Individual Transferable Quotas. National Academy Press, Washington, DC, 1999.

Copes, P., A Critical Review of the Individual Transferable Quota as a Device in Fisheries

Management, Land Economics, 62(1986):278-291.

Dupont, D., Individual transferable vessel quotas and e¢cient restructuring of the primary

harvesting sector, Annals of Operations Research, 94(2000):275-294.

Färe, R., and S. Grosskopf. Theory and Application of Directional Distance Functions,

Journal of Productivity Analysis, 13(2000):93-103.

Färe, R., S. Grosskopf and C. A. K. Lovell. Production Frontiers, Cambridge University

Press, Cambridge, UK, 1994.

Freid, H. O., S. S. Schmidt and S. Yaisawarng, Incorporating the Operating Environment

Into a Nonparametric Measure of Technical E¢ciency, Journal of Productivity Analysis,

12(1999):249-267.

32

Gauvin, J. R., J. M. Ward, and E. E. Burgess. Description and Evaluation of the wreck…sh

…shery under Individual Transferable Quotas. Marine Resource Economics, 9(1994):99-118.

Geen G. and M. Nayer. Individual Transferable Quotas in the Southern Blue…n Tuna

…shery: an Economic Appraisal, Marine Resource Economics, 5(1988):365-387.

Kirkley, J. and D. Squires, Capacity and Capacity Utilization in Fishing Industries, mimeo,

University of California, San Diego 1999.

Montgomery, W. D. Markets in Licenses and E¢cient Pollution Control Programs, Journal

of Economic Theory 5(1972):395-418.

Simar, L. and P. W. Wilson, Statistical Inference in Nonparametric Frontier Models: The

State of the Art, Journal of Productivity Analysis, 13(2000):49-78.

Squires, D., M. Alauddin and J. Kirkley, Individual transferable quota markets and invest-

ment decisions in the …xed gear sable…sh industry, Journal of Environmental Economics

and Management, 27(1994):185-204.

Squires, D., and J. Kirkley, Individual transferable quotas in a multiproduct common prop-

erty industry, Canadian Journal of Economics, 24(1996): 318-342.

Squires, D., H. Campbell, S. Cunningham, C. Dewees, R. Q. Grafton, S. F. Herrick, Jr.,

J. Kirkley, S. Pescoe, K. Salvenes, B. Shallard, B. Turriss and N. Vestergaard, Individual

Transferable Quotas in Multispecies Fisheries, Marine Policy, 22(1998): 135-159.

Thomas, J. S., G. D. Johnson, C. M. Formechella, and C. Riordan. Perceived Social and

Economic E¤ects of Current Management Policies on Red Snapper Fishermen Operating

in the Gulf of Mexico” unpublished report University of Southern Alabama, 1993.

33

US Department of Labor, Bureau of Labor Statistics, Employment and Earnings, January,

1994.

Waters, J. R. An Economic Survey of Commercial Reef Fish Vessels in the U.S. Gulf of Mex-

ico. National Marine Fisheries Service, National Oceanic and Atmospheric Administration,

unpublished report, July, 1996.

Waters, J. R. Quota Management in the Commercial Red Snapper Fishery, Marine Resource

Economics, 16(2001):65-78.

Weninger, Q. and R. E. Just. An Analysis of Transition from Limited Entry to Transferable

Quota: Non-Marshallian Principles for Fisheries Management, Natural Resource Modeling,

10(Winter 1997): 53-83.

Weninger, Q. Assessing E¢ciency Gains from Individual Transferable Quotas; An Appli-

cation to the Mid-Atlantic Surf Clam and Ocean Quahog Fishery, American Journal of

Agricultural Economics, 80(1998): 750-764.

Weninger, Q. and R. E. Just. Firm Dynamics with Tradable Output Permits, American

Journal of Agricultural Economics, to appear.

34