Technological change and the Tragedy of the Commons: The Lofoten Fishery over Hundred and Thirty Years Rögnvaldur Hannesson Kjell Salvanes Dale Squires

Period covered: (open access ended in 1988) Fish stock data since 1900

The traditional product: fish dried in open air

Main issues Productivity development (last year of open access) Puzzle: Labor productivity lagged behind agriculture & industry Role of fish stock for productvity Growth in TFP > growth in labor (or capital) productivity if role of stock taken into account Open access dissipates gains from technical progess

Labor productivity (value) in the Lofoten fishery (9-years moving average) versus wages in agriculture and manufacturing

Labor productivity (annual and 9-years moving average) in the Lofoten fishery

Strong correlation stock-catch Also between boats and men—we can’t use both in a regression, with boats as proxies for capital

Fishermen per boat

Production function: Y catch per year, E effort (boats or fishermen), S stock An instantaneous function not really appropriate for discrete time modeling, but the Lofoten fishery is seasonal (~3 months) with the fish migrating gradually to the area Fish migrations

abConstantR2R2 Gill nets ** (4.51) ** (2.71) (0.33) 0.27 Long line ** (2.07) ** (5.68) (-0.33) 0.34 Hand line (1.59) ** (2.17) (0.36) 0.10 Purse seine ** (4.11) (1.84) (0.48) 0.83 Dan. seine ** (4.25) ** (2.46) (-0.79) 0.47 Note 0 < b < 1, constant reflects technical progress

Constant  ln(Y/L)  ln(Y/L) t-1 R2R2 All boats (-1.26) (0.08) ** (4.35) 0.14 Gill nets (-0.83) (0.87) ** (5.74) 0.23 Long line (-1.83) (-1.52) ** (3.81) 0.15 Hand line (-0.40) * (-2.29) ** (3.65) 0.18 Purse seine (-1.01) (1.70) (-1.43) 0.60 Dan. seine (0.08) (0.32) (-0.02) 0.00

Change in participation (no. boats) related to catch per fisherman lagged one year Use no. fishermen & catch per fisherman, both lagged one year, as instrumental variables Using catch value per fisherman gives poorer results than just catch per fisherman Could be because price depends on quantity

Perioda0a0 a1a1 R2R (0.48) (-4.83**) (0.09) ** (-4.11) (1.34) ** (-3.10) 0.16 In 1938 fishermen got exclusive rights to sell raw fish. Seems to have stabilized prices (as intended).

lnBoatslnStockD00D10D20D30D40D50D60D70ConstR2R2 Gill nets.9319 (2.56).3586 (3.28) (2.87) (4.17) -.45 (1.57) -.51 (1.40) -.02 (0.11) -.34 (1.91) -.19 (0.86).23 (1.29) 1.74 (0.78).66 Long line.7082 (3.16).6417 (4.66) -.96 (1.86) (2.26) -.97 (1.89) (2.30) -.50 (1.41) -.51 (2.41) -.20 (1.06).43 (2.98) 1.65 (1.88).86 Hand line.7685 (3.77).4138 (2.01) (3.74) (3.46) (0.01) -.12 (0.24).09 (0.21) -.16 (0.45).04 (0.14).36 (1.31).83 (0.64).86 Danish seine (5.26).7296 (3.28) -.53 (2.16).06 (0.33) -.69 (0.36).84 Somewhat uneven technological progress, if even, we’d see d00 < d10 <…< d70 < 0 completed by the 1970s Note 0 < b < 1, 0 < a ~ 1

lnLlnBlnSDhandlineDlonglinetConstantR2R (4.16**).2965 (2.03*).4393 (5.82**) (-7.74**).1200 (1.48).0233 (15.91**) (-2.79**) 0.87 Estimates of a 1 and a 2 with panel data Growth rate of TFP: Set a 1 = 0.65, a 2 = 0.35

Average G based on 9-years moving average of TFP TFP growth > labor productivity growth Normal technological progress, diluted by decline in stock GearGill netsLong linesHand linesDanish seine Average growth rate of labor productivity Average growth rate of TFP TFP-index for end year

Two possible reasons why TFP grows faster than LP: 1)Rising no fishermen per boat 2)Decline in fish stock

Major episodes: Motorization Low prices 1920s & 30s, technology reversion Raw fish marketing board 1938 Better technology & outside options after 1950