Peak-to-peak analysis is a relatively simple method to assess the capacity utilization of an industry over time. An advantage of peak-to-peak analysis is that it requires information on only one output measure and one input measure, and hence is suited to estimating capacity utilization with only Level 1 data (see Table A.1). Peak-to-peak analysis has been applied in fisheries by Ballard and Roberts (1977), Ballard and Blomo (1978) and Hsu (2003).
Peak-to-peak analysis is based on an underlying assumption that output is a function of the level of inputs and a technology trend, such that
(1)
where Yt is the output in time, t; a0 is a proportionality constant; Vt is a composite or aggregate index of inputs; and Tt is the technology trend that represents productivity change. An implicit assumption in the use of a composite index of inputs is that the technology displays constant returns to scale. That is, increasing all inputs will result in a proportional increase in output.
The level of technology is determined by the average rate of change in productivity between peak years, where productivity is given by Yt/Vt (i.e. average output per unit of input). The technology in any one year is thus
(2)
where m is the length of time from the previous peak year, and n is the length of time to the following peak year, and Tt-m is the level of technology at the previous peak (i.e. year m) equivalent to the average productivity (e.g. catch per unit of effort) in that period. The other term on the right hand side (i.e. the term inside the brackets) represents the cumulative change in productivity between the two peaks. This is added to the average productivity in the previous peak year (i.e. year m) to give an estimate of the average productivity of capacity in subsequent years.
An alternative way of estimating the level of technology between peaks is given by
(3)
where Yn/Vn is the average productivity in the upper peak and Ym/Vm is the average productivity in the lower peak. The term in the brackets represents the average change in productivity between the two peaks. Both approaches produce identical results.
Assuming the proportionality constant has a value of 1, the estimate of the level of technology is equivalent to the capacity level of productivity (i.e. Tt = Yt*/Vt, where Yt* is the capacity level of output). From this, the capacity level of production can be estimated from the product of the inputs and the capacity level of productivity, such that
Yt* = VtTt (4)
and capacity utilization can be estimated by
CUt = Yt*/Yt. (5)
A particular difficulty in interpreting the results of a peak-to-peak analysis in fisheries is that no consideration is given to changes in the stock level. Apparent changes in productivity may be due to either changes in technology (the underlying assumption of the technique) or changes in the stock level.
This problem may be particularly pertinent in developing fisheries, where catch rates may increase rapidly initially, with the main peak occurring in the middle of the time series. Subsequent declines in catch rates may reflect falling stock levels. However, if the main peak is used as the last peak in the series (all other years showing a steady decline), it is likely that the technique will over-estimate capacity output and under-estimate capacity utilization.
The problem can be minimized by including lower peaks rather than successively higher peaks as is generally used in other industries that do not rely upon a biological resource base.
Data on the artisanal fishing sector in Nigeria were used as an example of how peak-to-peak analysis can be used to estimate capacity. The data were derived from Amire (2003), and are presented in Table B.1.
Table B.1 - Nigerian artisanal fisheries productivity, 1976-1994
|
|
|
|
Average catch per: |
|
Year |
Canoes |
Fishers |
Production |
Canoe |
Fisher |
1976 |
134 337 |
413 832 |
327 561 |
2 438 |
0.792 |
1977 |
137 447 |
424 838 |
331 280 |
2 410 |
0.780 |
1978 |
138 447 |
425 298 |
336 138 |
2 431 |
0.790 |
1979 |
133 728 |
446 152 |
356 888 |
2 669 |
0.800 |
1980 |
133 723 |
459 065 |
274 158 |
2 050 |
0.597 |
1981 |
120 142 |
440 592 |
323 916 |
2 696 |
0.735 |
1982 |
105 239 |
416 959 |
377 683 |
3 589 |
0.906 |
1983 |
129 555 |
472 122 |
376 984 |
2 910 |
0.798 |
1984 |
109 638 |
342 219 |
246 784 |
2 251 |
0.721 |
1985 |
80 688 |
302 234 |
140 873 |
1 746 |
0.466 |
1986 |
77 134 |
408 927 |
160 169 |
2 077 |
0.392 |
1987 |
76 644 |
437 465 |
145 755 |
1 902 |
0.333 |
1988 |
77 144 |
447 850 |
185 181 |
2 400 |
0.413 |
1989 |
77 155 |
470 250 |
171 332 |
2 221 |
0.364 |
1990 |
76 981 |
452 187 |
170 459 |
2 214 |
0.377 |
1991 |
77 093 |
457 102 |
168 211 |
2.182 |
0.368 |
1992 |
77 076 |
459 847 |
184 407 |
2 393 |
0.401 |
1993 |
77 050 |
456 381 |
106 276 |
1 379 |
0.233 |
1994 |
77 073 |
457 775 |
124 117 |
1 610 |
0.271 |
Source: Amire (2003).
The choice of input may have an impact on the measure of capacity output and, consequently, capacity utilization. In the Nigerian artisanal fleet, the number of canoes active in the fishery had declined over time while the number of fishers remained relatively constant (a result of more fishers operating per canoe). Over the same period, motorization increased in the fishery from 8.7 percent in 1996 to 20.8 percent in 1994 (Amire, 2003). As a result, it would be expected that there was substantial technological change in the fishery. Developing a composite index of inputs in such a case is difficult without first estimating a production function and imposing constant returns to scale.
For purposes of illustration, capacity was assessed using both canoes and fishers (separately) for the input measure. From Table B.1, it can be seen that the peak productivity periods for both inputs were 1976, 1979, 1982, 1988 and 1992. These peaks also are apparent by graphing the catch per unit input series (Figure B.1).
Figure B.1 - Catch-per-unit input, Nigerian artisanal fleet
Table B.2 - Peak-to-peak analysis using canoes as input measure
Year (t) |
Canoes (Vt) |
Production (Yt) |
CPUE (Yt/Vt) |
Average technological changea |
Capacity CPUE (Tt) |
Capacity output (Yt*) |
Utilization rate (Yt/Yt*) |
1976 |
134 337 |
327 561 |
2 438 |
- |
2 438 |
327 561 |
100% |
1977 |
137 447 |
331 280 |
2 410 |
0.0768 |
2 515 |
345 701 |
96% |
1978 |
138 247 |
336 138 |
2 431 |
0.0768 |
2 592 |
358 330 |
94% |
1979 |
133 728 |
356 888 |
2 669 |
0.0768 |
2 669 |
356 888 |
100% |
1980 |
133 723 |
274 158 |
2 050 |
0.3067 |
2 975 |
397 885 |
69% |
1981 |
120 142 |
323 916 |
2 696 |
0.3067 |
3 282 |
394 321 |
82% |
1982 |
105 239 |
377 683 |
3 589 |
0.3067 |
3 589 |
377 683 |
100% |
1983 |
129 555 |
376 984 |
2 910 |
-0.1981 |
3 391 |
439 289 |
86% |
1984 |
109 638 |
246 784 |
2 251 |
-0.1981 |
3 193 |
350 041 |
71% |
1985 |
80 688 |
140 873 |
1 746 |
-0.1981 |
2 995 |
241 631 |
58% |
1986 |
77 134 |
160 169 |
2 077 |
-0.1981 |
2 797 |
215 711 |
15% |
1987 |
76 644 |
145 755 |
1 902 |
-0.1981 |
2 599 |
199 161 |
73% |
1988 |
77 144 |
185 181 |
2 400 |
-0.1981 |
2 400 |
185 181 |
100% |
1989 |
77 155 |
171 332 |
2 221 |
-0.0020 |
2 398 |
185 055 |
93% |
1990 |
76 981 |
170 459 |
2 214 |
-0.0020 |
2 396 |
184 485 |
92% |
1991 |
77 093 |
168 211 |
2 182 |
-0.0020 |
2 395 |
184 600 |
91% |
1992 |
77 076 |
184 407 |
2 393 |
-0.0020 |
2 393 |
184 407 |
100% |
1993 |
77 050 |
106 276 |
1 379 |
-0.0020 |
2 391 |
184 192 |
58% |
1994 |
77 073 |
124 117 |
1 610 |
-0.0020 |
2 389 |
184 094 |
67% |
Note: Peak years in bold, a) estimated by [(Yn/Vn)-(Ym/Vm)]/(n-m)
The analyses, undertaken in an Excel spreadsheet, are given in Tables B.2 and B.3 using canoes and fisher numbers respectively. Average technological change was estimated between the peak years (indicated in bold). For example, between 1976 and 1979, average productivity change was (2.669-2.438)/(4-1) = 0.0768.
Capacity CPUE is estimated by adding the average technological change to the preceding years value. Capacity output is estimated by multiplying the capacity CPUE by the input level. The utilization rate is estimated by dividing actual output by capacity output.
Table B.3 - Peak-to-peak analysis using number of fishers as input measure
Year (t) |
Fishers (Vt) |
Production (Yt) |
CPUE (Yt/Vt) |
Average technological changea |
Capacity CPUE (Tt) |
Capacity output (Yt*) |
Utilization rate (Yt/Yt*) |
1976 |
413 832 |
327 561 |
0.792 |
|
0.792 |
327 561 |
100% |
1977 |
424 838 |
33 1280 |
0.780 |
0.003 |
0.794 |
337 461 |
98% |
1978 |
425 298 |
336 138 |
0.790 |
0.003 |
0.797 |
339 016 |
99% |
1979 |
446 152 |
356 888 |
0.800 |
0.003 |
0.800 |
356 888 |
100% |
1980 |
459 065 |
274 158 |
0.597 |
0.035 |
0.835 |
383 419 |
72% |
1981 |
440 592 |
323 916 |
0.735 |
0.035 |
0.871 |
383 540 |
84% |
1982 |
416 959 |
377 683 |
0.906 |
0.035 |
0.906 |
377 683 |
100% |
1983 |
472 122 |
376 984 |
0.798 |
-0.082 |
0.824 |
388 911 |
97% |
1984 |
342 219 |
246 784 |
0.721 |
-0.082 |
0.742 |
253 823 |
97% |
1985 |
302 234 |
140 873 |
0.466 |
-0.082 |
0.660 |
199 368 |
71% |
1986 |
408 927 |
160 169 |
0.392 |
-0.082 |
0.578 |
236 194 |
68% |
1987 |
437 465 |
145 755 |
0.333 |
-0.082 |
0.496 |
216 782 |
67% |
1988 |
447 850 |
185 181 |
0.413 |
-0.082 |
0.413 |
185 181 |
100% |
1989 |
470 250 |
171 332 |
0.364 |
-0.003 |
0.410 |
192 977 |
89% |
1990 |
452 187 |
170 459 |
0.377 |
-0.003 |
0.407 |
184 155 |
93% |
1991 |
457 102 |
168 211 |
0.368 |
-0.003 |
0.404 |
184 731 |
91% |
1992 |
459 847 |
184 407 |
0.401 |
-0.003 |
0.401 |
184 407 |
100% |
1993 |
456 381 |
106 276 |
0.233 |
-0.003 |
0.398 |
181 594 |
59% |
1994 |
457 775 |
124 117 |
0.271 |
-0.003 |
0.395 |
180 722 |
69% |
Note: Peak years in bold, a) estimated by [(Yn/Vn)-(Ym/Vm)]/(n-m)
Despite differences in the input measure used, the estimated capacity output was fairly similar in both instances (Figure B.2a). The estimated capacity utilization in each year was also relatively similar (Figure B.2b).
Figure B.2 -a) estimated capacity and b) estimated capacity utilization