5. The purse-seine fisheries for tunas in the Indian Ocean and the Atlantic Ocean

For the Indian and Atlantic Oceans we consider the Russell (1985) measure because of its ease of estimation, and impose variable returns to scale.

5.1 Data and methodology

Data was sought from around the world on fishing activity for the Atlantic and Indian Ocean purse-seine fisheries for tuna. Contacts were also made with ICCAT and the IOTC to obtain data. The data were determined to be inadequate for estimating capacity. Subsequently, data were obtained from Pallares et al. (2003) and Pianet et al. (2003) on the Indian and Atlantic Ocean fisheries, respectively. These data, however, were highly aggregated and inadequate for estimating capacity on a nation-by-nation basis or by fishing mode (e.g. sets on floating objects vs. sets on unassociated schools). It was subsequently decided to estimate capacity using aggregate annual data on the catches of yellowfin, skipjack, bigeye, albacore and all other species combined, numbers of vessels, fishing days, searching days, carrying capacity, a weighted mean of GRT, using the mid-point of vessel tonnage classes, and number of sets. Data were then converted to a per-vessel basis by dividing by the number of vessels in each year. Data on the Atlantic fishery and Indian Ocean fisheries were available for 1991-2002 and 1981-2002, respectively (Tables 5.1 and 5.2).

Unfortunately, the data were extremely limited in number of observations and detail, which might be important variables for estimating capacity (e.g. fishing days and searching days on schools associated with floating objects and unassociated schools, or activities and summary statistics by nation). The number of observations was, in fact, too few to consider all inputs. Unlike statistics in which the required degrees of freedom are well established, there are no specific required degrees of freedom. It has been well established, however, that too few observations leads to problems in DEA because of its orientation to relative efficiency. A rough rule of thumb offered by Cooper, Seiford and Tone (2000) is that the degrees of freedom (n) for DEA should be as follows: n $ max {m × s, 3(m + s)}, where n is the number of observations; m is the number of outputs; and s is the number of inputs. For the two data sets on the Atlantic and Indian Ocean purse-seine fisheries, we have five outputs and up to five inputs (average GRT, fishing days, searching days, carrying capacity and number of sets). We should, thus, have a minimum of 30 observations (m × s = 25, and 3(5+5) = 30). It was subsequently decided to use only average GRT per vessel per year and fishing and searching days per vessel per year. The GRT was considered as a fixed factor (i.e. could not be easily changed), and fishing and searching days were considered to be variable factors.

TABLE 5.1
Data used to estimate capacity in the Atlantic tuna fishery

Source: Pianet et al. (2003)

TABLE 5.2
Data used to estimate capacity in the Indian Ocean tuna fishery

Source: Pallares et al. (2003)

TABLE 5.3
Reported and estimated capacity output (tonnes) for the Atlantic Ocean purse-seine fishery

In actuality, the DEA problem used to estimate capacity has only one factor of production (GRT). This is because capacity can be estimated without including the variable factors. The constraint introduced by 8 ensures unrestricted use of the variable factors, which is equivalent to excluding the variable factors from Problems [1] or [2]. We, nevertheless, have a potential problem with degrees of freedom relative to estimating capacity for the Atlantic Ocean purse-seine fishery.

Capacity on a per-vessel basis was estimated for both the Atlantic and Indian Ocean fleets and subsequently converted to total fleet activity by multiplying the per-vessel estimates of capacity by the number of vessels in each year. We stress that because of the limited degrees of freedom and the paucity of the data relative to detailed activities of the various nations and the modes of fishing, our estimates represent extreme lower-bound estimates of capacity for the Atlantic and Indian Ocean purse-seine fisheries.

5.2 Results

5.2.1 Overall levels of capacity in the tuna purse-seine fisheries of the Atlantic and Indian Oceans

Estimates of capacity output on a per vessel basis for the Atlantic and Indian Ocean purse-seine fisheries suggest that both fisheries have some degree of excess capacity for all species (Tables 5.3 and 5.4). The highest degree of excess capacity (i.e. capacity output minus observed output per vessel) occurred for skipjack and yellowfin for both fisheries, which also had the greatest landings of all four of the tuna species.

TABLE 5.4
Reported and estimated capacity output (tonnes) for the Indian Ocean purse-seine fishery

For the Atlantic Ocean fishery during 1991-2002 the highest level of excess capacity relative to all species occurred in 1997 (Figure 5.1); the highest level of excess capacity for the Indian Ocean fishery also occurred in 1997 (Figure 5.2). The reason for this is unknown, but it may be a result of management or environmental conditions.

The Atlantic Ocean purse-seine fishery had the capability of harvesting 84 596 tonnes of yellowfin, 99 708 tonnes of skipjack, 24 585 tonnes of bigeye, 1 305 tonnes of albacore and 3 216 tonnes of other species per year (Table 5.3). Alternatively, the fleet had the capability to harvest 213 408 tonnes of all species combined. In comparison, the fleet had a reported average annual harvest of 79 140 tonnes of yellowfin, 82 778 tonnes of skipjack, 19 319 tonnes of bigeye, 587 tonnes of albacore and 2 132 tonnes of other species; the reported average annual harvest between 1991 and 2002 was 183 955 tonnes of all species combined. There was not, however, excess capacity for all species in all years. There was no excess capacity for yellowfin in 1992, 2001 and 2002; none for skipjack in 1991, 1993 and 1999; none for bigeye in 1994 and 1999; none for albacore in 1992 and 1999; and none for other species in 1996, 1998 and 1999.

The overall greatest level of excess capacity occurred in the Indian Ocean purse-seine fishery (Table 5.4). The estimated average annual capacity output between 1981 and 2002 for the Indian Ocean fishery was 109 256 tonnes of yellowfin, 129 156 tonnes of skipjack, 20 880 tonnes of bigeye, 2 855 tonnes of albacore and 1 729 tonnes of other species; the reported average annual landings were, respectively, 86 532 tonnes of yellowfin, 100 372 tonnes of skipjack, 14 647 tonnes of bigeye, 973 tonnes of albacore and 502 tonnes of other species. The average annual capacity output for all species was estimated to equal 263 877 tonnes, whereas the reported average annual total output was 203 026 tonnes. There was no excess capacity for yellowfin in 1981, 1983, 1988 and 1995; none for skipjack in 1981, 1983, 1989 and 2002; none for bigeye in 1981, 1986, 1987 and 1999; none for albacore for 1981-1983 and 1992; and none for other species in all years except 1991, 1994, 1996-1997 and 1999-2001.

TABLE 5.5
Excess capacity and full-utilization levels of variable inputs per vessel in the Atlantic Ocean purse-seine fishery

5.2.2 The Atlantic Ocean fishery

In the Atlantic Ocean fishery, a vessel had, on average, the capability to harvest an additional 322 tonnes of skipjack and 104 tonnes of yellowfin per year (Table 5.5). The total average annual excess capacity per vessel between 1991 and 2002 was 560 tonnes. They could do this by operating efficiently and making small increases in their fishing and searching days (the average annual number of fishing and searching days per vessel for the Atlantic fleet between 1991 and 2002 were, respectively, 259 and 232 days; the average annual level of fishing and searching days per vessel required to produce the capacity output were, respectively, 274 and 248 days). In general, the Atlantic Ocean purse-seine fleet could realize capacity output mostly by improving its efficiency (Table 5.6). The measure of CU adjusted for TE is quite close to one for most species and years, which indicates that gains in output could come mostly from operating more efficiently. The non-parametric Kruskal-Wallis test was conducted to determine the equality of observed and full-utilization levels of fishing and searching days; the equality was rejected at the 5-percent level of significance for both fishing and searching days, which implies that producing the capacity output would require an increase in fishing and searching days. The CU values were quite low for other species and albacore, which is the likely reason why the observed number of fishing and searching days were not equivalent to the levels required to produce the capacity output.

TABLE 5.6
Capacity utilization in terms of ratio of observed and technically-efficient output levels to capacity output levels in the Atlantic Ocean purse-seine fishery

TABLE 5.7
Observed and full utilization fishing and searching days required to produce the capacity output in the Atlantic Ocean purse-seine fishery

In addition to improved efficiency in operations, the average annual capacity output for the fleet could be realized with only a very modest increase in fishing and searching days (Table 5.7). The analysis suggests that fishing days should be increased by a meagre 5.5 percent to realize the capacity output, and the number of days spent searching by the fleet should be increased by only 6.8 percent.

5.2.3 The Indian Ocean fishery

In the Indian Ocean fishery, a vessel had, on average, the capability to harvest an additional 504 tonnes of skipjack and 616 tonnes of yellowfin per year (Table 5.8), both of which are considerably greater than the levels of excess capacity for these two species in the Atlantic Ocean fishery. The total average annual excess capacity per vessel for 1981-2002 was 1 327 tonnes. Vessels could realize the capacity output mostly by operating efficiently and making small increases in their fishing and searching days (the average annual numbers of fishing and searching days per vessel for the Indian Ocean fleet for 1981-2002 were, respectively, 224 and 185 days; the average annual level of fishing and searching days per vessel required to produce the capacity output were, respectively, 242 and 199 days). In general, the Indian Ocean purse-seine fleet could realize capacity output mostly by improving its efficiency (Table 5.9). The measure of CU adjusted for TE is quite close to one for most species and years, which indicates that gains in output could come mostly from operating more efficiently. The Kruskal-Wallis test was again conducted to determine the equality of observed and full-utilization levels of fishing and search days in the Indian Ocean fishery; results of the test could not reject the equality of reported and full utilization fishing and searching days. In other words, based on the non-parametric analysis, we conclude that the number of fishing and searching days required to produce the capacity output is equal to the reported or actual number of fishing and searching days. The exception is 1982, when the CU values were extremely low for yellowfin (0.51) and skipjack (0.68). The number of fishing and searching days would have had to increase 83.2 and 62.4 percent, respectively. Alternatively, we conclude that the capacity output could be realized mostly by improvements in TE only. In contrast to the Atlantic Ocean fishery, the CU values were quite high for other species and albacore.

TABLE 5.8
Excess capacity and full-utilization levels of variable inputs per vessel in the Indian Ocean purse-seine fishery

TABLE 5.9
Capacity utilization in terms of ratio of observed and technically-efficient output levels to capacity output levels in the Indian Ocean purse-seine fishery

TABLE 5.10
Observed and full-utilization fishing and searching days required to produce the capacity output in the Indian Ocean purse-seine fishery

Although results from the Kruskal-Wallis test suggest that realizing the capacity output requires only improvements in TE, there is still the possibility that gains could be realized by very small increases in fishing and searching days (Table 5.10). The analyses suggest that fishing days should be increased by a meagre 7.4 percent to realize the capacity output, and the number of days spent searching by the fleet should be increased by only seven percent.

5.3 Summary and conclusions

Overall, it appears that there is excess capacity in the Atlantic and Indian Ocean purse-seine fisheries for tuna. The more serious level of excess capacity exists for the Indian Ocean fishery. It was determined that, on an annual basis, there was approximately 61 000 tonnes of excess capacity in the Indian Ocean fishery. In comparison, the Atlantic Ocean fishery had approximately 29 500 tonnes of excess harvesting capacity. Alternatively, if Indian and Atlantic Ocean vessels operated efficiently, fully utilized their variable inputs and harvested the average annual reported level of landings, fleet sizes could be reduced, respectively, from 40 to 31 (22.5 percent) in the Indian Ocean fishery and from 53 to 46 (13.2 percent) in the Atlantic Ocean fishery.

We stress that the estimates presented in this paper are extreme lower-bound estimates of capacity. The limited number of observations and inadequate information for considering different modes and nations' fishing activities limits the estimation of the frontier or piece-wise technology. Alternatively, if there are few observations for estimating the frontier, DEA will tend to recognize each firm as being technically efficient and operating at full capacity. In this case, the observed or reported output will equal the technically-efficient output level and the capacity output level.