capacity in english channel fisheries

In this paper, the effects of stock abundance, output prices, fuel prices and fleet size on the rate of capacity utilisation are examined for a range of UK fleet segments operating in the

Factors Affecting Capacity Utilisation

in English Channel Fisheries

Diana Tingley and Sean Pascoe

Centre for the Economics and Management of Aquatic Resources (CEMARE), University of Portsmouth, Boathouse No 6, College Road, HM Naval Base, Portsmouth, PO1 3LJ, UK e-mail: diana.tingley@port.ac.uk

Abstract The harvesting capacity of the European fishing fleet far exceeds the reproduc-tive potential of the resource base As a result, most European Union fisheries are both biologically and economically over-exploited A series of fleet-reduction policies have been introduced in order to bring the harvesting capacity in line with target output levels However, the existence of unutilised capacity may reduce the effectiveness of these schemes as the remaining vessels may increase their individual capacity utilisation (CU), thus offsetting the effects of fleet reduction In this paper, the effects of stock abundance, output prices, fuel prices and fleet size on the rate of capacity utilisation are examined for a range of UK fleet segments operating in the English Channel Estimates of CU are derived using data envelopment analysis Results indicate that the average beam trawl vessel, using existing physical inputs, could potentially increase its revenue by a further 50%, assuming current fish stock levels and unrestricted access to resources The average gill net vessel could similarly increase its output by 43%, scallop dredge by 28% and otter trawl by 14% The results suggest that changes

in stock abundance are the main factor affecting CU, with no significant trends being observed for the economic variables

Keywords: capacity utilisation; data envelopment analysis; Tobit regression fish-eries; English channel

JEL classifications: Q22, C61

1 Introduction

It has long been recognised that the ability of the fishing fleet to harvest fish in the waters under the jurisdiction of the European Union (EU) far exceeds the ability of the stock to regenerate As a consequence, a range of measures to constrain the out-put of the fleet has been prompted under the EU Common Fisheries Policy (CFP)

This study was undertaken as part of the Commission of the European Communities Fifth Framework Programme Research Project (QLK5-CT1999-005), ‘Measuring Capacity in Fish-ing Industries usFish-ing the Data Envelopment Analysis (DEA) Approach’, (2001–02)

These include output controls (e.g aggregate quotas) and input controls (e.g gear restrictions, licence limitations, etc.) Capacity-reduction schemes have been imple-mented under the EU’s Multi-Annual Guidance Programmes (MAGP) as part of the Structural Policy of the CFP These capacity-reduction schemes have attempted

to reduce the amount of physical inputs, namely fishing vessel engine power (kW), gross tonnage and days fished, in different Member State fleet segments with the aim of reducing fishing effort and thereby catch levels

In the UK, the EU’s MAGP has largely been implemented through a series of decommissioning schemes in which ‘vessel capacity units’ (VCUs), a proxy measure

of the harvesting capacity of the vessel based on engine power and vessel size, were bought out from the fishery at public expense During the period 1993–2001, around £82m was spent by UK Fisheries Departments on the decommissioning of

839 vessels accounting for approximately 40,000 gross tonnes and an engine power 160,000 kW This represents approximately 18% of gross tonnage and 25% of engine power present in the UK fleet in 2002.1 Implicit in these attempts to reduce fishing capacity through the removal of VCUs is the assumption of constant returns-to-scale (CRS) (i.e removing one set of inputs results in a proportional reduction in total output) However, if a fleet segment2 operated at less than full capacity, then this relationship may not necessarily hold, as the remaining operators may increase their individual rate of capacity utilisation (CU), offsetting the effects

of overall reduction in ‘capacity units’ Pascoe and Coglan (2000) demonstrated that the effectiveness of this programme of decommissioning may have been less than that expected for beam and otter trawlers in the English Channel as a result of dif-ferences in the efficiency of fishing vessels (such that the effective capacity removed

is less than the nominal capacity) If the reduction in fleet size also creates incen-tives to increase CU, then the effectiveness of such programmes will be further diminished It is therefore beneficial to managers to have an understanding of the level of CU occurring in a fleet, the capacity of which they wish to manage, as well

as the factors that influence this level

Capacity utilisation is a measure of the actual to potential output given the exist-ing level of fixed inputs and assumexist-ing normal operatexist-ing conditions (Johansen, 1968) The concept of measuring CU in fisheries has gained increasing prominence in the last decade, particularly in response to the United Nations Food and Agriculture Organisation (FAO) International Plan of Action for the Management of Fishing Capacity (FAO, 1999) which recognised the importance of achieving a balance between harvesting capacity and available fish resources Data envelopment analysis (DEA) provides a flexible and adaptive method for measuring the level of excess capacity in a fishing fleet and has been used to estimate CU in a variety of fisheries (e.g Pascoe et al., 2001a,b; Dupont et al., 2002; Felthoven, 2002; Kirkley et al., 2003; Tingley and Pascoe 2003; Tingley et al., 2003; Vestergaard et al., 2003; Walden

et al., 2003) These studies have focused upon quantifying levels of CU rather than determining the factors which affect CU This study contributes to the growing body

of knowledge regarding estimation of CU by analysing the factors which affect the

1

Excluding vessels <10 m in length and mussel dredgers, neither of which was eligible for the schemes

2

The term ‘fleet segment’ refers to a portion of a fishing fleet typically defined by gear type

or fishing area

level of CU, in particular, exploring the effect of economic incentives in addition to issues of fishing vessel location and crowding CU in a fishing fleet is likely to vary over time for a number of reasons Assuming rational profit-maximising behaviour, fishers will increase utilisation until the marginal benefit of the derived output is equal to the marginal cost of producing it Hence, factors that affect revenue or costs affect the level of CU For example, seasonal changes in stock abundance may affect the catch per unit effort, and hence revenue Changes in fish prices also affect the marginal revenue, whereas changes in fuel costs affect the marginal costs

In this paper, factors affecting the rate of CU of four UK fleet segments in the English Channel are examined Measures of CU are derived using DEA DEA is a non-parametric, comparative method that can be used to estimate production possi-bility frontiers and where individual vessels lie in relation to this frontier The com-parative process compares the amount of output each vessel generates with the relative amounts of input used Felthoven and Morrison Paul (2004) describe these approaches as ‘technological-economic’, as the derived relationships are based on observed data that implicitly reflect underlying economic decisions The effects of seasonality, fish price, fuel price and crowding on CU are estimated using Tobit analysis, and the implications of the results for the effectiveness of decommissioning schemes are considered

2 Measurement of Capacity Utilisation

The technical measurement of capacity of a fishing vessel can be described by calcu-lating its potential output given its fixed factors of production CU can be defined

in terms of the ratio of actual (current) to potential (capacity) output CU is meas-ured on a [0,1] scale, where a measure of CU < 1 implies that the vessel, if fully utilised, could produce more than its current levels By implication, the same level

of catch could therefore have been taken by a smaller fleet consisting only of vessels that were fully utilised Capacity under-utilisation therefore is an indicator of the existence of excess capacity in a fishery and the measure can be used to provide an indication of the extent of excess capacity

The DEA model is formulated as a linear programming (LP) model Following Fa¨re et al (1989, 1994), for a vessel catching a set of m different species from n inputs (where n2 a is the subset of fixed inputs and n 2 ^a is the subset of variable inputs), this is given by:

max h1; subject to:

h1y0;m
X

j

zjyj;m 8m;

X

j

zjxj;n
x0;n n2 a;

X

j

zj¼ 1;

zj 0;

ð1Þ

where h1is a scalar denoting the output of the target vessel (i.e j ¼ 0) that can be radially increased, yj,m is the output m produced by vessel j, xj,n is the amount of

input n (n2 a; where a is the set of fixed inputs) used by vessel j and zj is a weight-ing factor such that capacity output is the weighted sum of the output of other ves-sels in the data set, including itself The value of h1 is estimated for each vessel separately, with the target vessel’s outputs and inputs being denoted by y0,m and

x0,n, respectively Only the set of fixed inputs are considered (i.e n2 a), as the assumption is that variable inputs are unconstrained The constraintP

zj¼ 1 impo-ses variable returns-to-scale (URS) on the production technology (Fa¨re et al., 1989) The level of capacity output is given by y0;m¼ h1y0;m for each species This radial expansion assumes that the catch composition remains constant, but the overall level increases through increased variable input use This is a fairly realistic assump-tion in most fisheries but not necessarily in all industries.3 CU is given by 1/h1,

Random variations in catch are included in the measure of under-utilisation rather than as stochastic error As a result, the estimated CU of the average vessel may be biased downwards (and capacity output biased upwards) where ‘luck’ is a major factor contributing to the catch of vessels operating on the production possi-bility frontier Furthermore, the observed outputs may not be produced efficiently and hence some of the apparent capacity under-utilisation may be due to ineffi-ciency (i.e not producing the full potential given the level of both fixed and variable inputs) (Fa¨re et al., 1994) Inefficiency may reflect differences in managerial skill, local resource abundance or even local management constraints (Felthoven and Morrison Paul, 2004) Although these factors may be overcome or reduced in the longer term (e.g through experience or training, relocation to alternative fishing grounds, etc.), in the short term they are likely to be capacity-limiting factors Hence, the potential production would be different for identical vessels operating under differing resource and managerial conditions.4

By comparing the capacity output to the technically efficient level of output, the effects of inefficiency can be separated from capacity under-utilisation As both the technically efficient level of output and capacity output can be upwardly biased because of random variability in the data, the ratio of these

Lee, 2002)

The technically efficient level of output requires an estimate of technical efficiency for each vessel, and requires both variable and fixed inputs to be considered The DEA model for this is given as:

max h2; subject to:

3

Fa¨re et al (1994) also present a model in which output mixes can vary Although fishers could potentially change their output mix through changing gear types employed or changing fishing areas, their ability to do either of these is limited in a given season Unless the vessels have already been constructed as multi-gear vessels, then changing gear types requires consid-erable costs Further, the geographical areas of many fisheries are large, and changing fishing location may require a substantial increase in costs through either increased steaming time or relocation to an alternative port

4

This argument can be extended to many other industries For example, farms face different resource endowments, whereas managers of firms in all industries have different abilities that are fixed in the short term

j

zjyj;m 8m;

X

j

zjxj;n
x0;n 8n;

X

j

zj¼ 1;

zj 0;

ð2Þ

where h2is a scalar outcome denoting the production of each firm that can increase

by using inputs (both fixed and variable) in a technically efficient configuration In this case, both variable and fixed inputs are constrained to their current level and h2 represents the extent to which output can radially increase through using all inputs efficiently The technically efficient level of output (yTE) is defined as h2 multiplied

by observed output (y), whereas the technical efficiency of the firm is given by

TE¼ 1/h2 As the level of variable inputs is also constrained, h2£ h1and the tech-nically efficient level of output is less than or equal to the capacity level of output (i.e y

TE
y) Then CU*is estimated by:

TE ¼ð1=h1Þ ð1=h2Þ¼

h2

h1

As h1‡ h2‡ 1, CU £ CU*£ 1; i.e the unbiased measure of CU is greater than the original measure, but£1 The difference between the measures reflects the degree to which random variation and technical inefficiency affect the output levels of the dif-ferent vessels

A number of scale assumptions may be incorporated into the analysis As noted

decreasing returns to scale to exist in the data set) In contrast, excluding this con-straint implicitly imposes CRS, whileP

jzj< 1 imposes decreasing returns-to-scale

jzj> 1 imposes increasing returns-to-scale (IRS), and P

jzj£ 1 allows for non-increasing returns-to-scale (NIRS, i.e either CRS or DRS) (Fa¨re et al., 1989) There are generally a priori reasons to assume that fishing would be subject

to DRS,5although this creates particular problems when assessing CU or TE – an artefact of the technique Smaller vessels, particularly in the fishery examined, are generally operated by part-time fishers As a result, there would be an a priori expectation that their capacity would not be fully utilised Allowing VRS would, in many cases, result in these vessels appearing to be fully utilised and operating under conditions of increasing returns, resulting in an underestimate of capacity and an overestimate of CU and TE Imposing DRS could potentially result in distortions for vessels operating on the frontier, as they could no longer be compared just with themselves but are forced to be compared with larger vessels that may be less effi-cient This, in turn, may result in infeasibilities A solution to this problem is to impose NIRS, which produces the same results for all vessels below the frontier, and allows vessels on the frontier to be compared only with themselves This may

5

This assumption generally holds for most demersal fisheries (Pascoe and Robinson, 1998; Herrero and Pascoe, 2003), although there is some anecdotal evidence that pelagic fisheries are characterised by increasing returns to scale

result in an overestimation of capacity and an underestimation of CU if IRS does exist for part of the fleet In this study, it was considered that the potential bias resulting from an assumption of NIRS was less than that which would result from imposing VRS

3 The Fisheries of the English Channel

Fishing activity in the Channel is generally based upon six major gear types: otter trawl, beam trawl, scallop dredge, nets, pots and handline These activities have been further classified into a number of me´tiers6 based on gear use, target species and area fished (Te´tard et al., 1995) Approximately 4000 vessels operate within the English Channel, over half of which are UK vessels (Te´tard et al., 1995) with many working part-time The smaller vessels are generally multipurpose, operating with different gears over the year and, in some cases, using different gears in the same month Large vessels tend to use the same gear over time but change me´tier by altering fishing grounds (Dunn, 1999) In total, 92 species are landed by vessels operating in the English Channel with the majority of landed weight and value being made up of less than 30 species

The data sets used in this analysis are derived from records of trip-level logbook observations compiled by the Fisheries Department of the Department for the Envi-ronment, Food and Rural Affairs (DEFRA) The DEFRA database covered the activity of UK fleets in the Channel over the period 1993–1998 It was

disaggregat-ed into observations by me´tier and trip level data were aggregatdisaggregat-ed to provide monthly levels of output and effort by vessel over the period examined The final database used in the analysis comprised only vessels fishing the same main gear type for at least four months a year and in at least three of the six years being studied

As a result, the data set used to calculate CU*scores was substantially smaller than the full set of available data and was comprised of 20,250 observations based upon activity using six main gear types: otter trawl, beam trawl, scallop dredge, nets, pots and handline, encompassing around 20 separate me´tiers Of these 20 me´tiers, four were subsequently chosen for the extended Tobit analysis investigating the possible factors affecting levels of CU The Tobit analysis focused upon some 9000 obser-vations relating to otter trawl, beam trawl (offshore), scallop dredge and gill net (gadoid) me´tiers undertaken in the Western Channel area

A feature of DEA is that it can incorporate multiple outputs into the analysis The key outputs used in the CU*analysis were the revenues of each of the top five species (in terms of value) Revenues from the remaining species were aggregated into an ‘other’ category These revenues were inflated to 1998 values using a Fisher price index to remove the influence of price changes between periods on the output measure A different Fisher price index was estimated for each fleet segment, repre-senting the different combination of species in the catch The key variable input

6

A fishing me´tier describes a specific fishing activity undertaken in a defined fishing area, tar-geting certain species and using a particular gear type For example, the beam trawl me´tier analysed in this study is undertaken in the Western Channel ‘offshore’ segment and targets key flatfish (mainly sole and plaice) using beam trawl fishing gear Beam trawling for similar species, using the same gear, but in the Western Channel ‘inshore’ area would be classified as

a separate me´tier

used to determine the TE and CU*scores was the number of days fished, while the fixed inputs were ‘deck’ size (estimated as overall vessel length * breadth) and engine power (kW) Average monthly vessel inputs and outputs are shown in Table 1

Additional information was required for the analysis of factors affecting capacity utilisation Average real fish prices were derived from the composite real revenue

operating in each me´tier in each month Average national marine diesel prices were obtained from the UK’s Sea Fish Industry Authority Statistical Department Although these prices were not specific to the fleets examined, a time series of monthly prices for fuel along the English Channel was not available Given that fuel operates in a competitive market, it can be assumed that movements of fuel prices in the Channel would be similar to the national price levels

Other factors that were thought of as being potentially important in determining

CU*were boat size, total fishing activity (represented by the number of boats recor-ded in the data in each time period), month (representing seasonal changes in stock abundance), home port (representing main fishing locations) and change in home port over the period examined (representing a change in key fishing location) These data were derived from the logbook

4 Estimation of CU*Scores

The estimation of CU*scores for multipurpose, multi-me´tier fisheries such as those found in the English Channel requires some additional considerations (Tingley

et al., 2003) As the DEA process compares a number of vessels’ inputs and out-puts, and no stock level data are available for most of the species, it is necessary to analyse data relating to only one me´tier at a time Therefore only vessels carrying out the same activity, in the same area and targeting the same species are compared

Table 1 Summary data used in data envelopment analysis, by vessel by month

Me´tier

Fixed inputs (kW, m2)

Engine power (kW) 152.74 (57.14) 456.59 (260.95) 341.16 (201.32) 147.05 (91.04)

Variable inputs (days at sea)

Outputs (revenue)

Note: All me´tiers are in the ‘Western Channel’ area, the beam trawl me´tier relates to the ‘off-shore’ area and the gill net me´tier relates to ‘gadoid’ species as defined by the Department for the Environment, Food and Rural Affairs

Values are given as average (SD)

However, in a multipurpose, multi-me´tier fishery this is problematic; analysing one me´tier at a time may result in an underestimation of capacity utilisation for vessels that operate in more than one me´tier in the period examined if a restricted data set

is used For example, the analysis may be focused on beam trawl vessels fishing in the Western Channel, however a vessel fishing in this me´tier with a Newlyn home port may, in reality, spend part of the month fishing in the Western Channel area and the other part fishing just outside the defined Western Channel area Excluding catch data relating to catches made outside the area would underestimate the level

of CU This problem can be overcome by including an extra variable input (days fished in other me´tiers) and an extra output (revenues resulting from activity in any other me´tiers) into the data set representing activity in any other me´tiers in the time period being analysed (Table 1)

scores for the me´tiers, calculated for the entire 1993–1998 period, were between 70 and 88% (Table 2), and individual vessel scores ranged from negligible levels to 100% (Figure 1)

The average vessel in each me´tier is operating at less than full CU, with the aver-age otter trawler achieving a score of 0.88, whereas the averaver-age beam trawler achieved only 0.70 Results indicate that using existing physical inputs, the average beam trawl vessel could potentially increase its revenue by a further 50%, assum-ing current fish stock levels and unrestricted access to resources The average gill net vessel could similarly increase its output by 43%, scallop dredge by 28% and otter trawl by 14% The results are broadly consistent with other studies on CU (Felthoven, 2002)

5 Factors Influencing CU*Scores

The key objective of this paper is to assess the factors that affect CU in fisheries

As noted above, it would be expected that profit-maximising firms would tend to operate at the point where their marginal revenue equalled their marginal cost In reality, fishers are generally unable to exactly determine their output because of the stochastic nature of the production process A number of stochastic elements always affect fisheries and are nearly impossible to capture in a data format Such random events include exceptional (or otherwise) luck, weather, disease outbreaks, break-downs and unpredictable stock biomass changes Despite this, it is expected that planned output would generally be based on expected yield, prices and costs

Table 2 Key me´tiers used in the analysis and average unbiased capacity utilisation (CU)

Me´tier

No of vessels

No of obs Top five species caught by value (1993–1998)

Avg

CU*

Note: All me´tiers are in the ‘Western Channel’ area, the beam trawl me´tier relates to the ‘off-shore’ area and the gill net me´tier relates to ‘gadoid’ species as defined by the Department for the Environment, Food and Rural Affairs

Channel fisheries tend to be highly seasonal CU*scores may, on average, tend to

be higher in the peak catching months of a fishery Conversely, in the height of its season, a fishery may become overcrowded and this may have a negative effect on

CU Annual changes in stock levels or management, for example, over the 1993–

1998 period may also affect CU in fisheries The major commercial stocks in the English Channel are generally fully or over-exploited and hence dependent on the yearly spawning stock biomass which may vary from year to year (Dunn, 1999) The physical area from which vessels operate may also influence CU* scores if, for example, vessels are located in relative close proximity to good fishing grounds Regional stock abundance is likely to have two effects on productivity, both related

to a lower average catch per day First, the lower catch rates will manifest them-selves in terms of lower apparent TE These effects are removed during the

number of days profitable to fish – having a direct impact on CU.7Smaller, inshore vessels – many of which use static gears – may be more affected by the location from which they operate than larger vessels that are capable of longer trips and can therefore operate further from their home port (and hence less limited by local stock abundance)

Changes in expected revenue per unit effort arising from changes in the average price of fish, would also be expected to change the incentives to fish, and therefore affect CU It would be expected that higher average prices would lead to increased

CU Conversely, changes in the price of fuel affect the marginal cost of fishing, and hence would also be expected to affect the number of days spent fishing High fuel prices are likely to have more impact upon vessels using mobile gears than those

Figure 1 Distribution of capacity utilisation scores

7

This is a drawback with estimating capacity utilisation from physical, rather than economic, data Economic DEA models have been developed that take into account the additional benefits and costs from increasing capacity utilisation that reduces this problem However, these require data that were not available for this study

using static gear, and are expected to produce an inverse relationship with changes

in CU*scores

Excessive levels of fishing effort are thought to generate crowding externalities, which may reduce revenue per unit effort and hence CU These may occur either as

a result of the nature of fishers’ working grounds that had previously been worked

by another fisher, or by fishers being forced onto less productive fishing grounds Assuming diminishing returns to effort in a fishery, decreasing levels of effort would result in a less than proportional decrease in catch, and hence an increase in average catch per unit effort of the remaining vessels This would increase marginal reve-nues, and hence provide an incentive to increase CU Pascoe et al (2001b) found evidence in the Dutch beam trawl fishery that increased aggregate fishing effort reduced the catch rate, manifesting itself as a reduction in the TE of individual vessels

The effect of boat size on CU was also examined Boats <10 m overall length (which make up two thirds of the Channel fleet) are generally less restricted by reg-ulation than their larger counterparts As a result, it would be expected that CU would decrease with an increase in boat size

6 Tobit Regression Analysis

The CU*scores derived using the above methodology were regressed against factors assumed to have influenced them Tobit regression was chosen over ordinary least square (OLS) because of the limited nature of the dependent variable (i.e unbiased

CU scores range between 0 and 1)

A range of dummy and continuous independent variables were used in the Tobit regressions Dummy variables were included for each month to capture seasonality effects Year dummy variables were incorporated to capture specific events or annual changes impinging on the CU*scores The base month and year for the ana-lysis of all me´tiers were January and 1993, respectively

The registered administration area of each fishing vessel was included to examine the importance of geographical proximity to fishing grounds and significant changes

of home port were included as an indication of possible changes in vessel owner-ship There are five vessel administration areas along the English Channel,8 stretch-ing from Newlyn in the west, to Plymouth, Brixham, Poole and finally Haststretch-ings in the east Spread throughout these administration areas are around 80 home ports The base for the administrative region dummy variables was Newlyn as this repre-sented the area in which most observations were derived

Indexed continuous variables included in the regression data sets were the monthly average (real) price of fish (£/kg for the main species annually caught by that particular me´tier in 1998 prices) and national monthly price of marine fuel The total number of vessels active each month in the me´tier was included to provide

a measure of crowding Finally, the vessel’s engine power (kW) and boat size (over-all length * breadth) were also included

The panel data prepared for each of the four me´tiers were analysed using SHA-ZAM (White, 1978) in both the linear and log-linear forms The models were initially estimated using all explanatory variables Where dummy variables

(repre-8

Defined by the UK Department for the Environment, Food and Rural Affairs