11

Overcoming the Common Pool Problem through Voluntary Overcoming the Common Pool Problem through Voluntary Cooperation: The Rise and Fall of a Fishery CooperativeCooperation: The Rise and Fall of a Fishery Cooperative

Robert T. DeaconRobert T. Deacon††, Dominic P. Parker, Dominic P. Parker‡‡, and Christopher Costello, and Christopher Costello

University of California, Santa BarbaraUniversity of California, Santa Barbara† † Resources for the FutureResources for the Future‡ ‡ Montana State UniversityMontana State University

NBERNBER

University of OsloUniversity of Oslo

June 17, 2010June 17, 2010

22

CollaboratorsCollaborators(Costello is on the left)(Costello is on the left)

33

What happened in Chignik …

• In 2001 some license holders petition to form a voluntary

cooperative.

• Co-op is approved and assigned dedicated share of catch.

• Regulator separates seasons of co-op vs. independents.

• Co-op shares profits equally; appoints fleet manager to

coordinate.

• Independents go about business as usual.

• Independents file suit against co-op; declared illegal after

2004.

Other Alaska salmon fisheries operated under existing regulations

(catch limit and season closure) throughout this period.

44

Research questions

• Are there gains from coordinating effort in harvesting a shared

resource? What kinds of gains? How large?

• Is it possible to structure fishery ‘reform without losers’?

55

Fishery policy questions

• Why not just rely on ITQs?

Slow to gain adoption;

Dividing TAC is contentious;

May leave some gains on the table.

• Allow a self-selected co-op to form, with dedicated share of TAC.

Chignik Sockeye Salmon Cooperative

66

Related literature:

• Efficiency of ITQ systems: Boyce (JEEM, 1992); Hannesson

(2004); Grafton, et al. (JLE, 2000); Costello and Deacon (MRE,

2007); Linn, Singh and Weninger (2008).

• Fisheries cooperatives and associations: Knapp (2002);

Matulich, Sever and Inaba (MRE, 2001); Johnson and Libecap

(AER, 1985).

77

Step 1: Model

• Model harvesters’ behavior as a 2-stage entry game.

• Stage 1: To join or not to join . . . .?

• Stage 2: Co-op maximizes group’s profit;

Independent fishermen take independent

actions.

• Solve (backwards) for SP Nash equilibria

Step 2: Empirics

• Test model’s predictions.

• Estimate co-op’s efficiency effects.

• Examine winners and losers.

88

Model Preliminaries:

Within-season model only; no dynamics.

Stock size is exogenous.

Total allowed catch (TAC) is set by regulator, exogenous.

To Join or Not to Join . . . . .

Before season, each fisherman decides whether to join co-op or fish

independently.

Co-op members’ actions are decided by co-op manager to maximize co-

op profit; members share profits equally.

Independents’ actions are decided individually, taking regulator’s and

other independents’ actions as given.

99

Regulator’s Actions:

Separates TAC between groups in proportion to number who join co-op.

Divides stock between groups by setting separate, disjoint seasons.

Controls total catch of each group by closing their seasons.

Because stock migrates, each group fishes its own stock without

interference from other group.

How Fishing is Carried Out:

Catch function: )( ZEZFQ (2)

Q = catch, Z = stock, E = effort

(F is fraction of stock caught)

0F , 0F , 0)0( F and 1)( ZEF

Catch is homogeneous of degree 1 in stock and effort

1010

Where Fishing is Carried Out:

Available fishing locations dd i ,0 : ‘Inside’ (near spawning river & processors)

vs. ‘outside’.

Stock enters outside zone first and migrates toward inside.

Effort applied outside contacts stock first, achieves high catch per unit effort.

Effort applied inside contacts smaller stock, achieves lower catch per unit effort.

Fishing is more costly outside.

All effort applied at a given distance gets same catch / effort.

1111

Catch Per Unit Effort at Different Locations:

Applying E0 units of effort to the stock yields Q0 catch;

- catch per unit effort equals slope of ab. (Catch from outside effort.)

Additional ET -E0 units of subsequent effort will contact a smaller stock;

- catch per unit effort equals slope of bc. (Catch from inside effort.)

Effort, E

Z

Fig. 1. Catch effort function

ZEFZ

0E TE

0Q

TQ

a

b

c

Fig. 1. Catch effort function

ZEFZ

0E TE

0Q

TQ

1212

Fishing Effort:

Fisherman h’s effort = hhT .

h ≡ h’s skill, Th ≡ time h spends fishing.

Regulator limits h’s effort by limiting Th, the time available for fishing.

Effort for the group of co-op joiners is Jh

hhT

, where J is the set of joiners.

Effort for the group of independents is defined similarly.

1313

Fishing Cost:

What is the public input? Information on stock locations, fishing

infrastructure, etc. Examples given later.

Opportunity cost of time? h could be teaching economics at a local college

instead of fishing.

Fishing cost: hhhhhihh xTTxGdc (1)

distance fishing time

skill level opportunity

cost of time

public input

public input contribution

effort cost/effort

1414

Co-op’s Objective (Stage 2)

Catch per member is exogenous (set by regulator) and price of fish

is fixed.* Therefore the co-op seeks to minimize cost:

JJi

iiiiJi

JixJiTd

xTTxGdJii

)(min;,

, (4)

subject to dd i ,0 , Ji TT ,0 for all i, the regulator’s effort limit

(governed by TAC rule), and TTJ , where TJ is the length of co-

op’s season, T is the time the stock is available, and xJ is co-op’s

expenditure for public input.

* Remember that the co-op’s stock is proportional to the number of joiners, to be determined in stage 1.

1515

Co-op’s Stage 2 Policy Choice

Proposition 1: The co-op’s optimal policy requires that:

I. All active co-op members fish as close to port as possible;

II. Only members with low cost per unit effort ii apply effort, these

efficient members fish the entire time the season is open, and the

season is open for T periods, the maximum time the stock is

available;

III. Provision of the public input equates the co-op’s aggregate marginal

benefit to marginal cost, satisfying a Samuelson efficiency condition.

1616

Co-op’s Stage 2 Policy Choice (detail):

(ii) Write the objective as follows:

JiiJi i

iJi

xJiTd

xTxGdJii

)(min

;,

, (4)

The co-op concentrates fishing among members with low ii (low cost

per unit effort). To do this it assigns zero fishing time to high cost

members. This slows the rate of fishing and extends the season.

(iii) Public input provision equates the co-op’s aggregate marginal benefit

to marginal cost, satisfying a Samuelson efficiency condition.

1)( 1 JJ ZFxG (5)

1717

Independent Fisherman’s Objective (Stage 2)

hhhhhIi ihhhhxTd

xTTxGdTHhhh

,,max (6)

where H is the independent group’s catch per unit effort. Independent

h solves (6) subject to regulator’s season length constraint, taking as

given the regulator’s and other independents’ choices. H is the

independent group’s catch per unit effort; for simplicity, we ignore

dependence of H on h’s choices.

1818

Independents’ Stage 2 Nash Equilibrium Strategy Profile

Proposition 2: The strategy profile for independents requires that:

I. Fishing distance: Some or all independents may fish ‘outside’.

II. Public input: Free-rider equilibrium. (At most, the highest skill

independent contributes a privately optimal amount; zero provision is

‘likely’.)

III. Fishing season is open only a fraction of the time fish are available.

(Also, season length is inversely proportional to average skill level.)

1919

Independent Sector’s Distance Choices (detail):

Average, ),( ZEA , and marginal, ),( ZEM , catch per unit effort; is effort limit

implied by TAC constraint; Note: If all fish inside or all fish outside, catch per unit effort is )( IA .

Independent effort, E

Fig. 2 Marginal and average catch per unit effort I1

IA

IM

ZEA ,

ZEM ,

1M

2020

Independent Sector’s Distance Choices (detail):

If all fish outside and 1 ‘defects’ inside, defector’s catch is )( IM ; defection inside

is profitable if dMA II )()( . If dMA II )()( , all fish outside is a NE. If all fish inside and 1 ‘defects’ outside, defector’s catch is )1(M ; defection outside

is profitable if )()1( IAMd . If )()1( IAMd , all fish inside is NE.

If )()1()()( III AMdMA , some fish inside and others outside is a NE.

Independent effort, E

Fig. 2 Independent fisherman h’s catch per unit effort, depending on where other independents fish

I

a

b

1

IA

IM

ZEA ,

ZEM ,

1M

2121

Independent Sector’s Distance Choices (detail):

Mixed equilibrium case: When all independents fish outside, the profit per unit effort from fishing inside is at point c. Transferring successive units of effort inside causes insider profit per unit effort to increase from point c toward point a. At outside effort level E , profit from fishing inside and outside are equal, so no one has an incentive to deviate. Could be multiple equilibria.

M(E))

A(E))

c

Outside effort

a

1 dEA )(

Insider profit per unit effort

Outsider profit per unit effort

E

Fig. 3. A strategy profile in which some independents fish outside while others fish inside is a NE.

M(1))

b

d

2222

Independent Sector’s Season Length (detail)

Iii

I In

KnZFT

)(/

)(/1

.

n(I) is number of independent fishermen; n(K) is total number of licensed fishermen;

Independent’s season length is inversely proportional to its average skill level.

2323

Stage 1 Decision: Whether or Not to Join

Individuals choose by comparing NE stage-2 profits; Difference in choices depends on skill and time cost parameters. Assumption (to get clear prediction): As we move to higher (higher skill) fishermen, (time cost) does not increase more than proportionately. (This ensures that cost per unit effort falls as skill increases.)

2424

Overall Prediction (Proposition 3)

Proposition 3: A SPNE strategy profile has the following properties:

I. Stage 2 behavior satisfies Propositions 1 & 2.

II. The group fishing independently consists of highliners; more

precisely, all independents have skill levels greater than any co-op

member.

Rationale: Highliners choose not to share profits with less skilled fishermen; highliners are most skilled at race to fish.

Comment: Multiple equilibria are possible; an equilibrium with all joining the co-op is

possible.

2525

Stage 1 Prediction (detail):

Size of co-op (fishermen ordered by )

Fig. 4. Equilibrium co-op size

C

)(Kn1 e e+1

)( im

)( iC

Note: Solid line shows co-op profit per member when new members are added in order of skill, . Dashed line shows profit of marginal independent, when independent fishermen are added to fleet in decreasing order of skill, , read right to left.

2626

Characterizing Pareto-improving Catch Allocations

Assumed catch is allocated on basis of group membership, as in Chignik. We allow for disproportionate allocations between sectors and solve for an allocation rule that is Pareto improving. The generalized allocation rule is:

)()( KnJnZZ J

Where JZ is the co-op’s allocation and is a positive scalar. Actual rule

had 1 . Key: Those choosing to fish independently will earn the same profit

regardless of whether a co-op forms if equals

Ratio of average skill for those who join to the average skill of entire fleet.

)(/

)(/

Kn

Jn

ii

Jii

c

2727

Pareto-Improving Catch Allocations …

Effect of allocation rule on those who choose to fish independently: They win if c , lose if c , and are indifferent if c .

Effect of allocation rule on those who join co-op: If c , the most skilled joiners lose if the co-op is allowed to form;

Rationale: in the limit the most skilled joiner earns the same profit as the least skilled independent.

If c , a cooperative forms and all who join are made better off;

independents are indifferent. Let cL be the lowest for which a cooperative forms. For cL ,

all fishermen benefit from allowing a cooperative to form.

2828

Pareto-Improving Catch Allocations … Proposition 4 The formation of a self-selected cooperative has the following

distributional consequences:

(i) If cL the institutional design is Pareto improving – fishermen of

all skill levels are made weakly better off by allowing the cooperative

to form.

(ii) If c the institutional design is not Pareto improving – all would-be

independents and some would-be cooperative fishermen are made

worse off by allowing the cooperative to form.

(iii) If L then no cooperative forms.

2929

Step 2 Empirics: Facts About Chignik Fishery

• One of Alaska’s oldest commercial fisheries (since 1880s)

• Purse seine fishing, with ~100 participants

• Managed by limited entry and season closures since 1974

• Significant monopsony power; only 1 or 2 processors.

3030

How the Co-op Worked in Practice

• Joiners sign 1-year contracts before season starts.

• Some members fish (22 out of 77) and are paid for effort.

• Profits after paying fishers are split equally.

• Elected board of directors and appointed manager to allocate effort.

• Motivated partly by intention to deliver higher quality product.

3131

Map of Chignik Management Area on the Alaskan Peninsula

3232

Map of Chignik Bay and Near Vicinities

‘‘Inside’ locationsInside’ locations‘‘Outside’ locationsOutside’ locations

3333

Purse seinerPurse seiner Desiderata Desiderata

(fished for co-op)(fished for co-op)

3434

Purse seiner Purse seiner completingcompleting set (August 2008) set (August 2008)

3535

Deck operations on Desiderata(unskilled labor, obviously)

3636

Weir installed on Chignik R. each seasonfor counting escapement.

3737

Counting escapement

3838

Counting escapement

3939Counting escapement

4040

Step 2: Empirical Approach

• Compare outcomes in Chignik vs. other neighboring fisheries before & during

co-op years

• Compare outcomes for co-op vs. independent groups

• Outcome variables:

value of fishing permits

proportion of permits fished

fishing locations

season lengths (speed of fishing)

catch price

• Compare the attributes of joiners vs. independents

• Compile (anecdotal) info on public input provision

4141

Test for change in permit values & elimination of redundant effort

Panel Regressions of Permit value and Proportion of Active Permits Independent Variables

(1)

Y = permit value

(2)

Y = proportion of permits fished

Constant

7,962*

0.329*

Co-op Years t-statistic

48,814* (2.32)

-0.267* (3.69)

lbs harvested (000s) -0.051 (0.43)

-1.17e-07 (0.29)

Fixed Effects Year dummies Fishery dummies

Included Included

Included Included

Observations Adjusted R2

66 0.873

66 0.807

4242

-0.2

0.0

0.2

0.4

0.6

0.8

1.0

1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008

chignik f ishery

average across other f isheries

differenceco-op years

Figure 5Proportion of Permits Fished in Alaska's Purse Seine Fisheries

4343

Test for change in fishing location & season length

Table 2: Time-Series Regression Analysis of Inside Catch and Season Length Independent variables

(1)

Y = proportion of catch from inside

(2)

Y = number of days fished

Constant

0.773*

471.13

Co-op Years t-statistic

0.267* (3.48)

32.16* (3.66)

lbs harvested lbs harvested2

4.63e-07 ---

0.0004 ---

Year Year2

0.039 ---

-106.93* ---

Observations F-statistic Adjusted R2

38 6.14

0.664

26 1.987 0.528

4444

Table 4 Panel Regression of Gross Earnings Per Pound (in 2008 dollars) Independent Variables

Y = gross earnings per pound

Constant

0.526*

Co-op Policy t-statistic

0.214* (2.22)

Fishery-Wide TAC -1.17e-06* (2.14)

Fixed Effects Year Dummies Fishery Dummies

Included Included

Observations Adjusted R2

66 0.814

4545

Table 5 Comparison of Mean Catch Histories for Ranked and Sorted Clusters of Fishermen

# of Obs. Mean

Catch Share t-stat for diff. (abs. value)

Independents v. All Joiners

Independents 18 1.29 2.90* All co-op members Fishing vs. Non-fishing Co-op Joiners

78 1.00

Co-op members who fished 18 1.11 1.83* Co-op members who did not fish

59 0.90

4646

Evidence on co-op public inputs

• Precise distance and temporal control of effort, exploit tides;

• Centralized info on stock locations, dispatched effort;

• Coordinated effort in order to:

• match processor capacity (released live fish!)

• meet fishery manager’s goals;

• raise product quality (delivered live fish);

• Installed stationary ‘fixed leads’ (funnel) on migration route.

4747

Approximate position of fixed leads

‘Fixed leads’

4848

Table 7 Proportionate profit increase from allowing co-op to form

Increase in license value $48,814

Baseline license value $185,806

Coop operating horizon (years) 3 5 10 ∞

Proportionate profit gain () (r=.10)

0.83 0.60 0.40 0.26

Proportionate profit gain () (r=.07)

1.11 0.79 0.50 0.26

4949

Was the Catch Allocation Pareto-Improving?

Actual allocation rule: co-op’s catch share allocation equaled its members’ average historic catch share.

- Model indicates this would yield a knife-edge weak Pareto improvement.

- A slight mistake could make highliners worse off.

As co-op membership increased over time, the allocation rule moved against the remaining independents.

- Model indicates this would make them worse off.

- To get a weak Pareto improvement, average independents’ catch share would need to rise by 10%;

- Average independent’s catch share actually fell by 40%.

5050

The Catch Allocation …

Two highliners brought suit to outlaw the co-op, won on a technicality.

- If you profit from an Alaska fishing permit, you must actually do the fishing.

- Court agreed cooperative fishing was efficient.

- Legislation was introduced to change the law, but it died.

5151

Discussion & Conjectures

• Efficient to coordinate use of a shared input inside a ‘firm’ (e.g., co-op) vs.

across a market (ITQ);

• ITQ systems do not:

Encourage public good provision;

Discourage secondary races to catch ‘best’ fish;

Supported by evidence from NZ CSO organizations.

• Assigning a single co-op share may be easier than assigning many

individual ITQ shares.

• Allowing voluntary co-op allows the disgruntled to opt out.

• Importance of a TAC division that makes all better off.

5252

Thanks for the invitation!