Abstract
We analyze the effect of technological change on labor and total factor productivity in the Lofoten fishery, using detailed data for 130 years. Our findings support the important role of natural resources in productivity and improvements in welfare in natural resource–based industries. The total factor productivity has risen faster than labor productivity in the fishery, indicating that the considerable technological progress in this industry has to some extent been neutralized by the decline in the fish stock. Open access to the fish resource most probably led to this situation. (JEL Q22)
I. Introduction
A received wisdom in economics is that new knowledge and technological change are the only source of growth and improved welfare. How does this play out in industries based on common natural resources? The ongoing discussion about improvements in fishing technology threatening fish stocks indicates that the fruits of technological progress might not materialize in such industries.
Our point of departure is the puzzle raised by the low productivity of a resourcebased industry over more than a hundred years: the Lofoten cod fishery. This fishery, taking place on one of the world’s most productive spawning grounds, had several major occurrences of technological change over this period, and yet labor productivity remained low for a long time, lagging behind productivity development in agriculture and manufacturing over most of the period.1 How could this be?
Figure 1 shows the development of wages in agriculture and manufacturing, as well as the value of the catch per fisherman (9-year moving average). Except for two brief periods (around 1870 and in the 1940s), revenues per fisherman lagged behind wages in agriculture and manufacturing until the mid-1960s, when they took a big leap forward. Comparing the 1970s and 1980s with the 1860s, fishermen did better than keep pace with agriculture and manufacturing, even allowing for some decline in their share of revenues. Understanding this lack of improvement, even with important changes in technology, and then the big leap in productivity, is of interest not only for this industry but also for similar situations in other resource-based industries.
Indices of Wages in Agriculture and Manufacturing, and 9-Year Moving Average of Revenues Per Fisherman
Our main strategy for analyzing the effect of technological and institutional change on the development of productivity in the Lofoten fishery is to focus on the interaction between technological and institutional change in this fishery over 130 years and the natural resource on which it was based. The theory of open access natural resources tells us that they will be overexploited and that labor productivity of the associated industries will be on par with comparable occupations. It also tells us that people will flock into such industries as occupations of last resort if opportunities in the labor market dry up, a process that in the end is likely to be self-defeating and to make a negligible and possibly negative contribution to the total wealth of the economy. We will analyze this issue in the Lofoten fishery by estimating productivity measures of the fishery, especially comparing labor productivity to total factor productivity in order to identify the role of the natural resource.
The most likely explanation of the lack of productivity gain in the Lofoten fishery over a hundred years is the fact that this was an open access industry. And so it remained for the entire period that we analyze (1860-1988). There are two possible and not mutually exclusive reasons for the leap in productivity after World War II. Many minor changes in technology could have come together and begun working in unison (Clark 2007). Large boats, decked boats, engines, better equipment, and so forth may not have been enough separately, but when all of the changes that had taken place could be combined, as they were in the 1950s, progress in productivity was possible. The other explanation is the rapid economic growth in the Norwegian economy since the mid-twentieth century. This provided better outside opportunities and is likely to have absorbed those who were least skilled at fishing and had the least-effective boats. By that time fishing was no longer an occupation of last resort.
Even if an open access resource industry appears stagnant in terms of productivity, technology could still be progressing at a rate comparable to other industries, with its fruits rotting away, as it were, through overexploitation of the fish stock. To examine this we will calculate total factor productivity, with the fish stock as a factor of production in addition to labor and boats. This concept is analogous to total factor productivity in other industries with one exception: higher total factor productivity in the fishery does not necessarily make anyone better off. Total factor productivity might grow in the fishery, and yet the total production of fish might decline for any given input of factors under human control, due to a declining fish stock. In an ordinary manufacturing industry, rising total factor productivity is good news because production can be maintained despite less use of labor and capital, making it possible to share the value produced among fewer people working and provide a higher return to the owners of capital. In addition the labor released and the capital no longer needed could be used for other purposes, making the whole economy better off. But decline of fish stocks has no such effects. A declining fish stock does not mean that more fish become available for some purpose other than fishing; a declining fish stock is likely to put an upward pressure on the cost of each fish caught and will ultimately reduce the surplus growth of the stock, which sets the limits for sustainable fishing. Growth in total factor productivity in the fishery while fish stocks decline would indicate that the fishing technology is progressing well, but it would not necessarily result in larger or cheaper fish catches. Hence, taking the role of fish stocks for productivity into account could explain what the technological development in the industry really is, whether or not it is comparable to other industries, and why it might be that technological progress in the fishery would not translate into larger fish catches and higher wages for fishermen.2
The Lofoten fishery has a long history. It is mentioned in Egils Saga, which deals with events that took place in the 800s A.D. Later, dried fish (stockfish) from Lofoten was a major item of trade for the Hanseatic League and the reason for their trading post in Bergen, which lasted for hundreds of years. It was an important source of income—or a gamble to bet on—even for people from distant parts of Norway, and still attracts fishermen from afar. For hundreds of years it was by far the most important fishery in Norway and indeed one of the most important sectors of the economy. The fishery exploits the annual spawning migration of the northeast Arctic cod, which in winter comes from the Barents Sea and the Norwegian Sea toward the Norwegian coast to spawn, mainly around the Lofoten islands (see Figure 2). The fishery is seasonal, beginning in late January and ending in late April.
Spawning Mlgrations of the Northeast Arctic Cod (From Hylen, Nakken, and Nedreaas 2008)
II. Data
Since 1859, the year the Lofoten Law of 1857 took effect, a report on the Lofoten fishery has been published annually. These reports contain much information on the fishery and events related to it.3 Up to the 1930s the information became increasingly detailed. There are records of how many fishermen and boats participated in the fishery, from 1875 broken down on the gear types used, how much fish they caught, on by-products such as fish liver, on prices of fish and by-products, the weather, the number of ships coming to Lofoten to buy fish, and much else.4 After the 1930s the degree of detail in the reports declined gradually, and as of the early years of this century they are just a six-page summary with a few graphs and tables, whereas in the 1930s they would run to over a hundred pages. This poses certain problems in getting long, consistent time series of variables.
The data on catches and participation in the fishery have been collected from these annual reports. The participation is based on a census of fishermen and boats at a certain date about half way through the fishing season, when the fishery is close to its peak.5 Participation varies throughout the season, increasing gradually up to a peak and then falling off toward the end. It is highly likely that people left the fishery early if the fishery was not going well, and their decision to begin fishing, or whether to participate at all, is likely to have been influenced by news about how well the fishery was going. After the telegraph and, later, telephone came on the scene, the fishermen had access to up-to-date news about this. The participation measures thus are imperfect in that they say nothing about the intensity of participation; clearly the fishing effort by a thousand boats over four weeks is not the same as that of eight hundred over six weeks; in fact the latter would be greater, all else equal.
The Lofoten reports contain no data on how much capital was invested in the boats, and neither do they specify the size or other physical characteristics of the boats that might be used as proxies for the capital investment. From other sources we know, however, that even if most boats were small, there were considerable differences in size and equipment. Beginning in 1936, cost and earnings studies of the Lofoten fishery were carried out until the 1960s, when such studies were extended to the entire Norwegian fishing fleet. These studies reported the size distribution of the boats participating in the fishery. Over this 30-year period this did not change a great deal; the range was wide, from less than 20 to over 80 feet length, but most boats were between 20 and 50 feet.6
The data on the fish stock stretch back to 1900 and were obtained from the Institute of Marine Research in Bergen. These data are based on stock assessment models, which follow the age composition of the stock through time. Even if these methods are known to have their limits in providing up-to-date assessments of the stock, their accuracy for any given year improves as time goes by and more becomes known about the life history of the different year classes of fish. The data on the stock provide figures for the mature part (which is exploited by the Lofoten fishery) separately and are expressed in weight units. The fish mature at an age of six to seven years, and the spawning stock consists of several year classes, the oldest fish being over 10 years. The stock data for 1900-1912 were estimated using the catch per fisherman.7 While these data undoubtedly are less than perfect, they certainly are the best available. There is reason to expect the more recent data to be more accurate than earlier ones, due to improvements in stock assessment methodology.
III. Background and Descriptive Statistics
Major Developments
Traditionally the Lofoten fishery was pursued with hand lines, each fisherman holding a line with a sinker and a baited hook. In the 1700s new types of fishing gear came into use: gill nets and long lines left lying in the water overnight. Originally there was much opposition to the novel gears from those who used the traditional hand line. A petition was sent to the King in Copenhagen (Norway was at that time under the King of Denmark), begging for these new gear types to be banned. The King was apparently sympathetic to the fishermen’s arguments, and in 1744 he sent a letter to the governor of Nordland County, admonishing him to allow only the use of hand lines, which “since time immemorial have been used in the fishery by rich and poor alike.”8 Over time the once so novel gear types became traditional and are now referred to as such, but opposition to new types of fishing gear or new technology has been a recurrent theme in the Lofoten fishery, and so have petitions to ban their use. Sometimes they have met with success, such as for seines in the latter part of the nineteenth century and the purse seine in the 1950s, while at other times the new technology has carried the day, such as engines in the early 1900s and the Danish seine after 1959.9
In 1816, after the restoration of the Norwegian state, albeit in a union with Sweden, a law on the Lofoten fishery was enacted. This law went a long way toward establishing what lately has come to be known as territorial use rights (Christy 1983). The fishing banks were divided into areas belonging to the nearest fishing base on land and further subdivided into fields where the boats were allowed to fish. The allocation of the fishing fields was in the hands of local governing committees, usually headed by the owner of the onshore facilities, which the fishermen had to rent for accommodation and for drying the fish. In practice these fishing fields became an informal property of the owner of the onshore facilities, being rented out with the base on land, with good fields carrying a premium rent (Solhaug 1983).
Over time, dissatisfaction with these arrangements developed. One important contributing factor seems to have been a shift in fish migrations from the eastern part of Lofoten to the western part (Solhaug 1983), but such shifts occur from time to time, presumably depending on environmental factors. Being tied to one particular fishing place impeded pursuing the fish wherever they happened to be, and in the western area there was more space and less need for a spatial regulation. This resulted in a new law on the Lofoten fishery, enacted in 1857. This law did away with the previous spatial regulation and made it clear that anyone had the right to fish wherever he wanted, but allowed for division of the fishing grounds between different gear types if deemed necessary to avoid entanglements. The new law thus established the Lofoten fishery as an open access fishery, but subject to certain rules of conduct. The law established a regulatory authority for the Lofoten fishery, whose purpose was ensuring that the rules of conduct were followed, such as not leaving harbor until the morning signal had been given, not setting any fishing gear after a certain hour in the evening, and respecting the division of the fishing areas between the various gear types.
The law of1857 certainly resounded with the prevailing free enterprise spirit of the time. The Lofoten reports also made it quite clear that the purpose of the law was to increase the efficiency of the fishery by making it easier to follow the fish to wherever they happened to be and to allow the fishermen to use their skills and endurance to the utmost. This was long before overexploitation of fish resources had been recognized; on the contrary it was widely believed, even by prominent biologists, that fish resources were inexhaustible and that fishing made no difference whatever for their fecundity (Smith 1994).
Over time, discontent also developed over the new law. As the number of fishermen increased, spatial control again became necessary to avoid gear collisions. The law allowed for dividing the fishing banks between different types of gear, but not between boats, and the latter arrangement apparently was not what the fishermen sought. Instead they sought influence over the division of ocean space between different gears and other regulatory issues, not least because of the development of new gears such as seines, which most of them wanted banned, and because of attempts by some landowners to close off certain areas where the fish concentrated and either have it all for themselves or charge fishermen for access to these areas. Yet another law was enacted in 1897, giving fishermen influence over management through consultative committees, while keeping the ultimate authority and enforcement firmly in the hands of the state. This arrangement has basically prevailed until this day.
The Lofoten fishery remained an open access fishery until 1990. There were no restrictions on the number of boats, fishing time, or quantity of fish caught.10 In 1989 a limit was set for the first time on how much fish could be caught, and the fishery was stopped half way through the season when this limit had been reached. In the following years individual boat quotas or quotas for groups of boats were used, which limited the amount of fishing time at least some boats could spend in this fishery. Because of this our analysis of the fishery ends in 1988, the last year without regulations affecting fishing time and the quantity caught.
Technical and Institutional Change
In this section we will describe the development of participation, technological and institutional change taking place in different periods, as well as the development of catch and stock in the Lofoten fishery. The main aim is to identify different periods in the fishery in terms of technology and participation. Figure 3 shows the development of men, boats, catch, and fish stock from 1860 until 1988. Participation in the Lofoten fishery, whether measured as number of fishermen or boats, evolved in a somewhat cyclical fashion; it was high in the 1880s and 1890s and peaked again in the 1930s. It declined sharply from the early 1950s to 1965 and continued declining after that, but at a lower rate and somewhat irregularly.
Fixed Base Indices for Fish Catch, The Number of Fishermen, Fishing Boats, and Spawning Stock in the Lofoten Fishery 1860 1988
Based on Figure 3 and the somewhat scarce literature, we think five different periods can be distinguished (Brandal 1981, 1982; Solhaug 1983; Lefdalsnes 1969; Iversen 1937; Johansen 1999; Martinussen 2006; Mathisen 1981; Vea 1984):
Until about 1880: mostly small and open boats without sail, using gill nets, long line, and hand line.
From about 1880 to 1906: Open boats were replaced by decked boats with sail, even if many open boats remained. Decked boats were larger than the open boats. Sail boats used small boats (“dories”) for nets and long lines.
From 1906 to 1920: Motorization of the fleet. By 1920 most of the decked boats had engines. Most of the change took place from about 1912 to 1920.
From 1920 to about 1940: Two distinct changes occurred: (a) mass increase in the participation of open boats using hand line and long line; (b) enactment in 1938 of a law giving fishermen monopoly of sales of fish at the first hand (the so-called Raw Fish Law).
From the mid-to late 1950s: introduction of nets made of synthetic fiber, and, later, development of increasingly sophisticated jigging machines for hand lines. Probably also general improvement in technology proceeding in small, incremental steps.
Participation in the Fishery
The long-term changes in participation (Figure 3) are probably related to both demographic development and “outside options.” Before 1900 there was substantial immigration to Nordland County, but substantial emigration, both to the rest of the country and to America during the two decades after 1900. In the two decades from 1920 to 1940 the emigration to America slowed to a trickle because of restrictions in the United States, and emigration to the rest of Norway also declined because of a slowdown in industrial development and economic growth in general.11 After 1950 the Norwegian economy entered a phase of full employment and rapid economic growth, coinciding with a rapid decline in the participation in the Lofoten fishery.
The variations in participation are rather moderate compared with the fish catches, which display both short-term and longterm variability. Much of this variability is evidently related to variations in the fish stock. Very roughly we can identify two periods of large but variable catches: 1860 to 1900 and mid-1920s to 1950. From 1900 to the mid-1920s and after 1950 catches were much lower and less variable, except for a peak in the early 1970s.
Changing Gear Type
Figure 4 shows the number of boats using the three traditional gear types: gill nets, long lines, and hand lines. A conspicuous feature is the enormous bulge in the number of hand line boats from 1920 to 1965, rising from a few hundred in 1920 to over 5,000 in the early 1930s and then dropping off gradually, with some fluctuations. The surge in the number ofhand line boats was accompanied by an increase in the number of fishermen per boat, so for a time that number was about the same as for the long line boats (see Figure 6), while both before and after it was much lower. The increase in the number of hand line boats in the 1920s was not due to switching gears; the number of long line boats was also increasing while the number of gill net boats changed little. It thus appears that many owners of relatively large boats decided to participate in the Lofoten fishery in the 1920s and then gradually pulled out (or switched gear) from about 1930 onward. The number of long line and gill net boats has also been variable, but less so, especially the gill net boats. There is some evidence of gear-switching between gill net and long line boats up to about 1920 in that we find significantly negative correlations for some subperiods.
Number of Boats Using Nets, Hand Lines, and Long Lines
It is very tempting to see the increase in the number ofhand line boats in the 1920s as a result of poor outside opportunities together with continuing population growth (Lefsdalsnes 1969). The 1920s were a period of economic stagnation in Norway, and the emigration to the United States had declined substantially. Hand lines were the cheapest fishing gear available and so the poor man’s choice. What is a bit surprising is that this development was to some extent reversed in the 1930s, also difficult times with few opportunities. Some of that may be due to frustrated expectations; the Lofoten reports in the 1930s make references to fishermen leaving Lofoten poorer than when they came and in need of financial help to go back home.
Labor Productivity
Figure 5 shows the development of average labor productivity, measured as catch volume and value per fisherman. There was surprisingly little trend in productivity until the mid-1960s, when it was in fact little higher in volume terms than a hundred years before. Because of the enormous interannual variability, moving averages provide a better basis of comparison than comparing just initial and final years. Figure 5 also shows 9-year moving averages, which give a reasonable smoothing of the curves. In volume terms, labor productivity was almost three times higher in the 1970s and 1980s than in the 1860s, but in terms of real value it was 15 times greater. Given that the Norwegian economy grew impressively over the said period, fishermen would have been in dire straits indeed if rising prices had not come to the rescue.
Fixed-Base Indices of Productivity (Catch Per Fisherman), In Value and Volume Terms, Annual Values, and 9-Year Moving Averages (Value Index Has a Constant Money Value)
This rise in the revenue per fisherman coincides with a decline in the number of fishermen per boat, as seen from Figure 6. Another noteworthy feature is that rising prices compensated for the fall in labor productivity in the 1940s and 50s. In this period, as well as later, prices were regulated by the Raw Fish Marketing Board.
Fishermen Per Boat
Even if labor productivity in volume terms remained fairly constant for a hundred years, there was some variation in the number of fishermen per boat, as is seen from Figure 6. We can identify four phases. From 1860 to about 1900 the number of fishermen per boat increased for all three gear types. In the early 1900s the number fell significantly over a relatively short period, first for gill net boats, and later for the long line and hand line boats. The Lofoten reports mention two reasons for the decline in the number of men per boat in gill netting in the latter half of the 1890s.
First, beginning in the early 1890s, winches were gradually introduced to pull the nets. This, according to the same reports, did not result in a smaller crew until a few years later, however. Second, from 1895 a new type of net, the so-called small fish net, came into use. This required a smaller crew than the other type of net.
That winches did not immediately result in smaller crews indicates that it takes time for a technology to diffuse and for fishing firms to adjust to the new technology. We see a similar pattern in the motorization of the boats. Although first introduced around 1900, this process did not take off until about 1915, but had largely run its course in 1920, even if many open boats without engines still remained.
The first mention of engines in the Lofoten reports occurred in 1895, when a gill net boat was reported to have used a petroleum engine, which apparently did not work very well. Over the period 1907-1915 much of the fleet was motorized, although boats without motor would be in use for many years after that.12 This period coincided with a decline in the number of men per boat in the long line and the hand line fishery. Replacement of oars and sails by engines is therefore likely to have been labor saving, although the Lofoten reports make no mention of this. Shortly after 1910 the number of men per boat began to increase again, first for the gill net boats and later for the long line and hand line boats. This could possibly be because boats propelled by engines instead of sails and oars could be made bigger, accommodating a larger crew. Then, finally, since about 1950 the number of fishermen per boat has fallen almost uniformly. This period coincides with strong economic growth and growing real wages in the Norwegian economy, which is likely to have encouraged substitution of capital for labor, but only after 1965 did labor productivity increase substantially in volume terms (Figure 5).13
IV. Productivity and Fish Abundance
Technical Change
A commonly used production function for the fishery is
[1]where t is an index for time (year), Y is the amount of fish caught, E is fishing effort, S is the fish stock, and A reflects the technology.14 Since we have data on both men and boats, we could in principle split effort into these two components, but the correlation between the two is so high that it is not meaningful to include both in a regression for each gear type individually.15 First we shall report results from using boats as proxy for effort (using men instead gives qualitatively similar results) and then consider results for a panel of the gear types represented in the entire period, using both men and boats as regressors.
We assume that technological progress is of the Hicks-neutral type and hence will be manifested in rising values of A over time. To estimate this, we use time dummies for periods of equal length. This gives us the following equation to be estimated:16
[2]where the d’s are dummies for the T periods into which we divide the entire period, dropping the dummy for the last period.
Estimating [2] without time dummies is plagued by strong serial correlation; the Durbin-Watson statistic is below or close to 1 (see Table 1). Even with time dummies, serial correlation persists to some extent. We deal with this by using the Prais-Winsten procedure. As to the time dummies, we first split the 89 years into two periods of 44 and 43, years, then split into four periods of 22 except for the last of 23, and so on, finally settling for 15 periods of 6 years, except for the last one of 5.
Results from Estimating Equation [2] by Successively Including Regressors
As the number of time dummies is increased, they will probably start picking up some of the influence of the stock (and possibly effort as well), which varies over time in an irregular, but wavelike fashion. This is likely to result in a drop in b, as less of the variability over time will be explained by the variations in the stock. We find that as we go from time dummies representing 11 years to dummies representing 6, b in fact increases slightly for long lines and Danish seine (from 0.73 and 0.90 to 0.76 and 0.95), while it falls for gill nets and hand lines (from 0.51 and 0.61 to 0.41 and 0.54). Furthermore, the time profile of the dummies also gets a bit jagged as we go from 11 to 6 years. As the results for the 11-year dummies are similar to the 6-year ones and a finer resolution is preferable, we report the results for the 6-year dummies.
The parameter estimates are shown in Table 1. We report results of three regressions where we increase the number of regressors successively, from (1) effort (boats) to (2) effort and fish stock, and finally (3) effort, fish stock, and time dummies. Using only effort as a regressor results in a significantly positive estimate of a, but only for hand line and Danish seine do we get coefficients close to one, the value we would expect on the basis of constant returns to effort. For long lines and hand lines we get values significantly below one. When also including the fish stock as a regressor we get a significantly positive estimate of b for all gear types except hand line, while the estimate of a is not greatly affected except for long line, which is about halved, but still significantly positive. The popular Schaefer production function implies b = 1, but we get this only (or nearly so) for Danish seine. For the three “traditional” gear types we get values below one. Such values are not entirely unreasonable; they would result if a small stock occupies a smaller area than a large one, and the fishermen know where to find it.
Results from Estimating Equation [2′] By Successively Including Regressors
Including the time dummies as regressors gives an estimate of a close to one for gill nets and Danish seine, and a value that is significantly positive but also significantly less than one for long lines and hand lines. The estimates of b are all significantly positive but significantly less than one, except for Danish seine. The pattern of the time dummies, normalized with respect to the last period, is shown in Figure 7. The pattern is broadly similar for all gear types; the overall impression is a rising trend, but with setbacks.
The Pattern of Time Dummies (Equation [2], CF. Tables 1 and 2)
Pooling the data for the three gear types represented for the entire period, it is possible to estimate separate coefficients for the effort variables (fishermen and boats), albeit not with great precision. The equation to be estimated becomes
[2′]where L denotes the number of fishermen, B the number of boats, and dHL and dLL are dummies for long lines and hand lines, respectively. The results are shown in Table 2. The dummy for hand line boats is significantly negative, indicating that they are less technically efficient than long line and gill net boats. This is as expected, since hand line boats are smaller and employ fewer people. The dummy for long line boats is not significant, so their technical efficiency is indistinguishable from gill net boats, also as expected. The time dummies reflecting the technology show a positive and somewhat more even trend than for the individual gear types (Figure 7), and their significance is much improved. Roughly speaking it seems possible to identify four major periods:
From the beginning of the twentieth century to the end of World War II. This period is characterized by a rather steady technological progress, especially for gill nets (annualized rate about 4%). The motorization of the fleet, which took place about 1905-1920, is a likely reason for the rising productivity in the first part of this period. For hand lines there is a particularly rapid growth of productivity from about 1915 to the early 1920s, but this is probably due to large boats converting from nets and long lines to hand lines, rather than technological progress; as Figure 4 shows, the number of hand line boats increased enormously from about 1920 to 1933, and the historical literature mentions hand lines as the only affordable gear in the 1920s because of hard times in the fishery (Figure 6 shows an increase in the number of fishermen per hand line boat in the early 1920s). For long line boats the productivity growth is sluggish; there is actually a decline from about 1910 to about 1920, and the annualized growth rate for the entire period is less than 1%.
From the end of World War II to about 1960. This period is characterized by stagnation or decline for all three gears. Some of this could be explained by hard times in the immediate postwar years; there was rationing of foreign currency for imported equipment, and the fleet was obsolete after insufficient renewal during the war. We would not expect this to have lasted so long, however; the Norwegian economy picked up relatively quickly, and the 1950s were a period of unprecedented growth.
The 1960s to the early 1980s. In the 1960s technology improvement seems to have been particularly rapid for the three traditional gear types, with an annualized rate of growth of about 10% for all three. This strongly positive trend can be explained by the general economic progress in Norway during this period, except that we would have expected to see signs of it earlier.17 The boats became better equipped and fewer. It is likely that the least effective ones disappeared; this would show up as an improvement in technology. Some technological innovations were introduced shortly before this period; the most important one was probably synthetic fibers in nets and lines, which came in the mid-1950s. Technology improvement appears to have run its course by the 1970s, except for the Danish seine, which came on the scene in 1959 and for which technology seems to have improved continuously until the early 1980s. The stagnation that appears to have occurred for the other gears after 1970 is surprising. It is true that synthetic fibers had become universal by then, but we would still expect to see traces of improved equipment. In particular we know that jigging machines for hand lines were introduced in the 1970s and 1980s and became more and more effective, but there is no trace to be seen of this.
From the early 1980s to the end of open access in 1988. This period shows an abrupt fall in productivity, for reasons we do not understand. The resource base was under an increasing pressure, and to deal with that the activities of the fishing fleet came to be restricted, but such regulations did not set in until 1989. Perhaps as fine a time resolution as 6 years is picking up interannual noise.
Total Factor Productivity
The estimates of the a coefficients in Table 2, with the time dummies included, are a1 = 0.80 and a2 = 0.12. The sum ofthis is just slightly below one. From cost and earnings studies of the fishing fleet we would expect the share of labor to be of the order 70% to 80%, so the estimate of a1 is not unreasonable. Using factor shares we can calculate the growth in total factor productivity (TFP) and compare with the growth in labor productivity. The growth rate of TFP is
[3]Using a1 = 0.8 and the b estimates from Table 1 Regression (iii) for the individual gears gives the results reported in Table 3. These results are based on 9-year moving averages of TFP and average labor productivity, as using beginning and end years would give an undue weight to these years, which in fact were atypical. It can be argued that a labor share of 0.8 is quite high, except for the labor-intensive hand line, but reducing this share to a possibly more reasonable 70% would not qualitatively change the results. According to Table 3, the growth rate ofTFP has been 2% to 2.5% per year, which probably is comparable to other industries in Norway. Hence, technological progress in the fishery has probably been comparable to that in other industries. The growth in average labor productivity has been slower, or 1% to 1.8% per year, which can be attributed to a decline in the fish stock over the entire period. The table also shows the growth rate in TFP including only labor and boats but excluding the fish stock. This results in a rate of TFP growth that is lower than for average labor productivity alone, except for Danish seine, where it is slightly higher. Hence, accounting for the influence of the fish stock shows a higher growth of TFP than leaving it out.
Rates of Growth of Total Factor Productivity and Labor Productivity, Based on 9-Year Moving Average
Further Econometric Issues
Several points can be raised concerning the above estimation procedure. Two important ones are errors in variables and simultaneity bias. Errors in variables are likely to be most serious with respect to the fish stock variable, which is an estimate taken from fish stock assessment models. We tried using remaining stock (stock less catch) lagged one year as an instrument for the stock, but a Hausman test showed that this procedure is not better than ordinary least squares estimation (see Table A1 in the Appendix).18 As to simultaneity, it can be argued that fishing effort is an endogenous variable, determined by expectations about the fishery, which are likely to be based on the results the previous fishing season. We tried using lagged effort and catch per fisherman as instruments for effort, but a Hausman test for this also indicated that this procedure is not better than ordinary least squares estimation (see Table A2 in the Appendix). This is not entirely surprising; when using the lagged variable as the instrument, basically the same information is used, due to strong serial correlation, so it is not expected that very different results are obtained. Other reasonable instruments were not available to us.
V. Conclusion
For about a hundred years labor productivity in the Lofoten fishery varied a great deal but hardly increased, while from the mid-1960s to the 1980s it almost trebled. Still, this would not have been enough to prevent the fishery from lagging behind agriculture and manufacturing; but with prices rising still more rapidly, revenues per fisherman in the end did better than keeping up with the wages in these sectors, using 1860 as a base year. Productivity in value terms (constant value of money) began to outpace physical productivity in the late 1930s, at which time price regulation was introduced. This may have resulted in higher prices and seems to have stabilized prices, making them less dependent on the quantity caught. From the mid-1950s price subsidies were provided, which is likely to have led to higher and more stable prices. These subsidies were not eliminated until after the end of the period being analyzed here (Hannesson 1996).
It is puzzling how the Lofoten fishery could continue for about a hundred years without any increase in labor productivity, while wages in other sectors of the economy rose. The answer may lie in the fact that the Lofoten fishery is a seasonal fishery, combined with participation in other fisheries and agriculture. The returns from these activities may have been sufficient to justify participation in the Lofoten fishery, especially since this coincided with low activity in the other occupations. Furthermore, the Lofoten fishery provided cash income, while the farming activities with which it was combined were largely for subsistence. This may have meant an extra premium for the Lofoten fishery, even if the results were meager. Many fishermen did, however, combine the Lofoten fishery with other fisheries that also provided cash income.
From the early 1950s the participation in the Lofoten fishery, both in terms of the number of fishermen and the number of boats, fell steeply. This was a time when the Norwegian economy reached full employment and attained a high rate of economic growth. In the mid-1960s and possibly earlier, labor productivity (in value terms) in the Lofoten fishery began to outpace wages in agriculture and manufacturing, and from the mid-1960s the decline in the participation in the Lofoten fishery was slowed considerably, but not reversed. The productivity improvement in the Lofoten fishery was largely due to rising prices; in terms of volume it still lagged behind wages in agriculture and manufacturing. It is possible to argue that the rise in prices soothed the impact of open access and kept people in the fishery to a greater extent than warranted.
Total factor productivity has risen faster than labor productivity in the Lofoten fishery. While the technological progress in the fishery has probably been comparable to that in other industries in the long term, even if it has been uneven over time, it has to some extent been neutralized by a decline in the fish stock. This decline would have prevented the Lofoten fishermen from reaping the full benefits of technological progress, but rising prices came to the rescue.
It is tempting to ascribe the slow development in labor productivity over almost a hundred years to the classic open access curse. As discussed above, the Lofoten Law of 1857 established the Lofoten fishery as an open access one and did away with whatever exclusive use rights that had developed in the fishery prior to that time. Improving the lot of fishermen was the most likely intention; this was well before the concept of biological overfishing had been generally recognized, and the science of fisheries biology was yet to be developed.19 In the longer term, however, this may have been self-defeating through encouraging excessive participation in the fishery and exposing it to the familiar evils of open access.
There are three types of problems caused by open access that we may identify for the purpose of this discussion:
Excessive fishing, leading to a diminishing fish stock and lower future returns
Overcrowding on the fishing banks, with one fisherman’s gear taking fish that another could have taken later in the season
Overcrowding so that one fisherman’s gear intervenes directly with another’s, for example, through entanglements
The first two of these are resource stock externalities, while the last is a pure crowding externality. As to the first effect, this was most likely present to some extent. The mature part of the cod stock consists of several year-classes of fish. Increased fishing effort in any one year lowers the survival rate of the year-classes, which reduces the stock of fish available in later years. The number of boats and fishermen was at its highest in the two decades before 1900 and again in the 1920s and 1930s, and so the stock externality is likely to have been most prominent in those years. These were also periods of declining labor productivity, preceded by a period of higher productivity that may have called forth greater fishing effort. But each year the mature stock is replenished with new cohorts. This was increasingly determined by what was left over by the ocean fisheries that developed over the last century and mainly fished young, immature fish. Until the early 1930s the catches in the Lofoten fishery were in some years almost one-half of the catches of the northeast Arctic cod, but after 1950 they seldom exceeded 10%. As the twentieth century proceeded, the decline in the fish stock available for the Lofoten fishery was increasingly due to a rising fishing pressure in the offshore fisheries.
The Lofoten reports in the 1930s make repeated references to excessive participation in the fishery, alleging that the revenues per fisherman would most likely improve if there were fewer of them. It was never made explicit exactly why this would happen: there is no reference to a “dynamic stock externality” by which excessive fishing in one year would reduce the catches in the coming years. By the 1930s it was certainly well known that the Lofoten fishery was based on several year-classes and that the survival rate of the fish to a large extent determined how much would be available the next year; in fact, there are supplementary chapters on this in many of the annual reports in the 1930s and even earlier, contributed by fisheries biologists. It is likely, however, that those who penned the annual reports had in mind problems 2 and 3 above, namely, conflict between gears or gear types on the fishing banks, which increasingly had to be divided between different gear types to avoid conflicts, and fish being caught early in the season instead of being available later if there had been fewer fishermen and boats. With respect to the latter, it does not seem to have been serious enough to affect the time pattern of landings.
In the difficult years of the 1930s, new and more productive technology was the solution envisaged in the Lofoten reports. Experimenting with new types of gear had a long history in the Lofoten fishery, and so did the opposition to any such or other novelty. The opposition in the 1700s to gill nets and long line was mentioned earlier. The Lofoten reports in the 1800s inform that some fishermen believed that the underwater telegraph cables killed the fish; dead fish were reported to be deposited along the tracks of these cables (how did people know?). The report writers in the 1800s promoted scheduled calls by steamships to outlying parts of Lofoten during the fishing season, but some fishermen thought this would scare the fish away. As the boats became equipped with engines, the sound from these was alleged to do so. Experiments with seines were made in the late 1800s, but in the 1890s all seines were banned from Lofoten, due to pressure from fishermen. From the 1920s on government-financed experiments with various types of seines took place, and around 1950 the Lofoten reports were enthusiastic about the purse seine. In 1950 this gear began to be used on an appreciable scale, but it took only a couple of years for the reports to express great doubts about this. The fishermen using the “traditional” gear saw it as a threat, in the spirit of their forerunners in the 1700s who felt the same way about gill nets and long lines. The purse seine took up a lot of space, and it was quite effective, threatening to make a large number of fishermen redundant. In 1958 it was outlawed.
The “long waves” in participation in the Lofoten fishery are consistent with the role of an open access fishery as livelihood oflast resort, where people enter or exit as a result of demographic pressures or rises and falls in outside options. The immigration to Nordland County before 1900 was accompanied by increasing participation in the fishery, the emigration, both to America and to the rest of the country, coincided with a decline in participation, and participation in the fishery reached new heights as emigration, both to America and the rest of the country, more or less came to an end in the 1920s and 1930s. Then, as the Norwegian economy entered the era of full employment and economic growth after World War II, participation in the fishery declined, especially as regards the number of fishermen, and labor productivity in the fishery rose.
The history of the Lofoten fishery has a familiar ring to it; this is not unlike the situation in the fisheries of many of the world’s poor countries today. For a long time the Lofoten fishery played the role of employment opportunity of last resort; when times were bad people resorted to it, because of its open access character. There was political pressure from fishermen to keep things that way, and pressure to ban new technologies that would save labor in the fishery and would be expensive for fishermen to acquire. Open access and prohibition of some efficient technologies (but not all) may have succeeded in distributing incomes more widely, but one may ask whether increased participation led to lower returns and only succeeded in distributing poverty more widely. Banning more efficient technologies and the development of access rights in suitable localities is certainly likely to have meant foregoing increases in productivity.
And as times got better, the exodus began. Productivity improved, partly through technological improvement but probably because of less participation as well. The productivity improvement may well have been self-enhancing, due to a positive feedback on technology; new technology is typically expensive and its returns uncertain, so poor fishermen are unlikely to introduce it, as is also abundantly clear from the opposition to new methods that always has come from the less well off fishermen.
Appendix Hausman Tests of Errors in Variables and Misspecification
Test of Errors in Effort Variable, Using Stock Remaining After Fishing Previous Season as Instrument for Stock
Test of Model Misspecification, Using Effort and Catch per Fisherman in Previous Season as Instruments for Effort
Footnotes
The authors are, respectively, professor, Department of Economics, The Norwegian School of Economics and Business Administration; professor, Department of Economics, The Norwegian School of Economics, and Statistics Norway Center for Economics Education 9CEP) and IZA; and senior scientist, Southwest Fisheries Center, La Jolla, California. We thank Ola Grytten, The Norwegian School of Economics and Business Administration, for access to data on wages for the manufacturing and agricultural industry in Norway.
↵1 We do not have absolute values for productivity in agriculture and industry, only indices. Hence, what we can state unambiguously is that productivity grew more slowly in the Lofoten fishery than in the two other industries we compare with.
↵2 There are not many productivity analyses that take the role of the fish stock into account. Squires (1992) pioneered the use of growth accounting in this context. Follows-up of this are provided by Fox et al. (2003) and Hannesson (2007). Jin et al. (2002) made a thorough study of productivity in the New England fishery, utilizing 30 years of data on fishing boats and fish stocks.
↵3 Since 1906 these reports have been published by the Directorate ofFisheries (Fiskeridirektoratef), but prior to that by its forerunner (Norges Fiskeristyrelse) and the Department of the Interior (Departementet for det Indre). All the reports are available electronically from the Institute of Marine Research, Bergen.
↵4 From 1879 there are detailed tables on the diseases treated by the resident doctors during the fishing season, from which we can find, among other things, how many persons were treated for syphilis (the maximum was 14 in 1895, and the last case was recorded in 1952) and how many teethwere pulled (a maximum of 1,346 in 1935). We can also find how many tradesmen and other professionals of various kinds traveled to Lofoten during the fishing season to offer their services. One such was quacksalvers, of whom there was a maximum of four in 1898, but none in some years, and they dropped out entirely after 1935.
↵5 Before 1918 this was March 16, but March 22 from that year on.
↵6 These reports were published in the series Årsberetning Vedkommende Norges Fiskerier, and later in the journal Fiskets Gang, published by the Directorate of Fisheries (Fiskeridirektoratet), Bergen.
↵7 For details on the data, see Hylen (2002).
↵8 The King’s letter refers to some of the arguments advanced against the new gear types. They were said to be too expensive for the common man to acquire and to cause much distress to Our Lord through encouraging the use of foul language and fights. The letter is quoted by Steen (1930, 30).
↵9 Polanyi (1944) makes this type of point repeatedly when discussing the transformation to market economy in Great Britain. Introduction of new technology and widening and deepening of market economy could be very disruptive to society, and at times the king would try to slow down the pace of change, always a rear guard action, but this allowed society time to adjust. So, this is not a unique story.
↵10 There were a few exceptions to this. In 1981-1984 the fishery was stopped in-season for about two weeks, which included the Easter holiday, during which little fishing would have taken place anyway. In the 1950s and 1960s it occasionally happened that the fishery was halted for a few days or limits were set on the catches of individual boats because of market problems.
↵11 On the demographic development up to 1940 and the importance of“outside options,” see Vea (1981).
↵12 That this transition was ongoing but incomplete by 1910 is clear from the Lofoten report for that year. This was apparently not a good year, and many fishermen blamed the noise from motorized boats for having scared away the fish. The report concurred with these boats having been numerous and the noise formidable, but was otherwise out of sympathy with the complainants, considering the argument as being vicarious for arguing against competing and better equipped fishermen.
↵13 The Lofoten reports are surprisingly silent on what technical changes may have been involved. One change that apparently increased productivity, although probably not of a labor-saving kind, was the use of synthetic materials in nets and lines. This transition took place in the mid-1950s.
↵14 This is a stock-flow technology; at a point in time a flow of effort is applied to the existing stock of fish to produce a flow of output.
↵15 The correlation between boats and fishermen ranges from 0.93 to 0.99.
↵16 This is inspired by Baltagi and Griffin (1988) and Baltagi, Griffin, and Rich (1995), but due to insufficiently large cross section and problems of multicollinearity we have been forced to use time dummies representing several years instead of annual dummies.
↵17 Gradual diffusion and learning by doing could account for this late improvement in technology due to innovations in the mid-1950s, but it is our impression that the synthetic fibers spread rapidly and that little adjustment was needed to use them effectively.
↵18 The cod is a long-lived species, and so the stock available each year is determined by the surviving stock from the previous season and recruitment of new, mature year-classes.
↵19 The history of the science of fisheries biology is told by Smith (1994).
