Estimating Elasticity of Demand for Fresh Bluefin Tuna in the World’s Largest Fish Market Kanae Tokunaga * Abstract This study estimates the demand elasticity for Bluefin tuna in the Japanese market by using an instrumental variables approach. Variability in the catch by purse seine vessels and fluctuations in the auctioned volume at Tsukiji Market are used as supply shocks. In 2014, the International Union for the Conservation of Nature placed Pacific Bluefin tuna the Red List following a sharp decline in its stock. This led to an increased pressure to reduce catch. This study aims to show how a catch reduction would impact prices. This study finds an elastic demand, which may be problematic because a catch reduction would not be compensated by an increase in price. 1 Introduction Japan is the largest consumer of tuna in the world. Over a quarter of global tuna catch is consumed in Japan annually (OPRT, 2004). Among all tuna species, Bluefin tuna has the highest price per kilogram. It is reported that the Japanese market accounts for 80% of the global Bluefin tuna trade (McCurry, 2015). In recent years, a growing concern for Bluefin tuna stock triggered a discussion about tightening the regulations for Bluefin tuna fisheries. Pacific Bluefin tuna stock was placed on the Red List by the International Union for Conservation of Nature (IUCN) following an assessment that concluded that the stock is threatened and vulnerable to extinction (IUCN, 2014). In 2015, the Fishery Agency of Japan introduced a catch limit on juvenile tunas (i.e. individuals that weighs less than 30 kilograms). While currently, there is no limit posted on the catch of matured Bluefin tuna, there is an increasing pressure to decrease the catch in order to conserve the stock. * [email protected] 1 This study estimates the price elasticity of demand for fresh Bluefin tuna in the Japanese market. There is a growing interest in understanding the implications of fishery management on the seafood market. Yet, the limited availability of market data and the complexity of fishery harvesting behaviors pose a challenge to conducting thorough analysis. In this study, we adopt the instrumental variables approach and use auction data from the Tsukiji Central Wholesale Market. Unlike agricultural commodities such as grains and legumes, storability is limited for fresh seafood products. While tunas are often frozen or canned, these products are treated differently from fresh products in the market. Because Bluefins are usually consumed raw, the landed tunas are usually transported to the wholesale market within a day. All of these reasons combined make fresh Bluefin tuna the ideal candidate for the application of instrumental variables to estimate demand. Among other existing studies that estimate the price elasticity of demand by applying the instrumental variables approach, the most notable and relevant to this paper is Angrist et al. (2000). The study uses weather as an instrument for supply shocks to measure the elasticity of demand for Whitefish in the Fulton Fish Market in New York. The use of weather and climate variables are perhaps more common in the study of agricultural commodities, as described by Angrist and Krueger (2001). Weather variables, such as precipitation and wind speed, serve as good instruments for those fish species that are caught by methods sensitive to weather conditions. However, they are not ideal for the estimation of Bluefin tuna demand for two reasons. First, purse seine vessels, which account for a significant amount of Bluefin catch within Japanese waters, are large enough to remain in the water for an average of two weeks. Because of this, the operation’s sensitivity to weather conditions is difficult to measure. Second, Bluefin tuna fisheries in Japan operate at various locations and times of year, using various types of gears. No single domestic fishery operates year-round. This is because tuna is highly migratory, and localized fisher folks can only target Bluefin tuna on a seasonal basis. These two characteristics invalidate weather conditions as a potential instrument for calculating demand elasticity. Instead, this study uses two types of supply shocks for the estimation of the price elasticity of demand for fresh Bluefin tuna. First, intra- and inter-annual variabilities in purse seine landings are used as quasi-experiments that shift fresh Bluefin tuna supply. Purse seine Bluefin tuna fishing takes place in the summer months. The landed volume by the purse seine fishery is the largest among all Bluefin tuna fisheries in Japan. Judging by the fact that 50% of the domestic catch quota for juvnile Bluefin is alloted for the purse seine fishery, their impacts on the market are significant. Inter-annual variations in purse seine landings are also considered. Although there is no conclusive study that explains the 2 causes of the variability, it is believed to be due to oceanographic conditions which cause schools of Bluefin tuna to disperse. This makes it difficult for purse seine vessels to harvest, which leads to lower landings in the years that the oceanographic conditions persist. We also use daily and monthly fluctuations as supply shocks. We rely on the characteristics of the purse seine Bluefin tuna fishery to form a quasi-experiment that shifts supply without affecting demand. Second, fluctuations in daily and monthly auctioned volume at Tsukiji Market are used as a supply shock. To do this, the deviation from the time trend is calculated from daily and monthly auction data. Because landed fresh Bluefin tuna is auctioned within a day, fluctuations in landings are reflected in the daily auctioned volume at Tsukiji Market. This characteristic makes the fluctuations in auctioned volume a suitable variable for supply shocks. Our method for calculating supply shocks is partly motivated by Roberts and Schlenker (2013), which estimated elasticities for agricultural commodities. They calculate yield shocks by deriving deviations from the time trend that is approximated by a restricted cubic spline. We adopt their approach to derive daily and monthly landing shocks. There are alternative methods to estimating demand elasticities for fish. A review by Asche et al. (2007) indicates the systems of equations approach, such as the Almost Ideal Demand System (AIDS) by Deaton and Muellbauer (1980) or the Rotterdam System by Theil (1965) and Barten (1968), to be a common approach. In an analysis of the Japanese seafood market, Eales et al. (1997) finds an inelastic price response to a change in consumption by using the AIDS model. In addition, retail demand for tunas in Japan is found to be inelastic by Wessells and Wilen (1994). While previous studies that calculate elasticities of demand do not distinguish between different species of tuna, this study distinctively analyzes the Bluefin tuna market. While there is evidence for substitutability among tropical tuna species, such as that of Yellowfin and Bigeye tuna, as presented by the market cointegration analysis of Bose and McIlgorm (1996), to the author’s knowledge, the substitutability between Bluefin tuna and other tunas is not yet fully understood. It is beyond the scope of this paper to test for market cointegration of different species of tunas. Rather, this paper focuses on understanding fresh Bluefin tuna. At Tsukiji Market, different species of tunas are auctioned separately and treated as different products. Fresh and frozen products are also auctioned separately and treated as different products. This indicates there exists a distinct market for fresh Bluefin tuna. In addition, management measures are usually implemented at a species level. Hence, a distinction between different species of tuna during market analysis adds value. 3 2 Background: Bluefin Tuna Fisheries in Japan In Japan, coastal fishermen target Bluefin tuna by using various types of gears at various locations. The main gears used are longline, trolling, setnet, and purse seine. Coastal fishing in Japan is managed by either region-based fishery cooperative associations or by gear-based associations (Makino and Matsuda, 2005). All of the purse seine vessels that target Bluefin tuna belong to the purse seine gear association and land in Sakaiminato. Other gears are managed by local fishery cooperative associations. Because Bluefin tuna is a migratory species, no fishery targets Bluefin tuna year-round. While almost all the Bluefin tuna that are caught in the coastal and off-shore waters of Japan are sold fresh, the catch from distant waters is sold either fresh or frozen. A report by World Wildlife Fund Japan indicates that the majority of fresh Bluefin are sold via wholesale markets, while frozen ones are sold directly to supermarket chains and trading companies without going through wholesale markets1 . Since 2004, the Bluefin tuna catch by purse seine vessels in the Sea of Japan has increased dramatically. Purse seiners that target Bluefins operate in the Sea of Japan by targeting those that congregate for spawning. The identified spawning area is depicted in Figure 1 in dark blue. Although Pacific Bluefin tuna migrate throughout the Pacific Ocean, they spawn in just two areas, the larger of which is located near Taiwan and the Okinawa Islands, and the smaller of which is located in the Sea of Japan. The spawning of Bluefin tuna in the Sea of Japan takes place in July and August. The fishing season for purse seine vessels starts at the end of May and lasts until July. From 2011 to 2015, 11 purse seine vessels landed their catch in Sakaiminato Port. In short, approximately half of the supply of fresh Bluefin caught in Japanese domestic waters is landed by a small number of vessels at Sakaiminato Port in about a 70-day period each year. Bluefin tuna fishing by purse seine vessels started in 1982 (Tottori Prefecture Fisheries Research Institute, 1995). Because purse seine vessels can only target fish that swim in a large group, Bluefin was not considered a good target by purse seine fishermen until a purse seine vessel targetting sardines and mackerel found a congregation of Bluefin tunas in the Sea of Japan, and later, scientific research found a spawning area nearby. Prior to this discovery, the majority of Bluefin tuna were caught by longline, pole-and-line and set-net. As a result of the discovery, Bluefin tuna became a primary target in the summer months when they congregate to spawn in the Sea of Japan. 1 WWF Japan (in Japanese) http://www.wwf.or.jp/activities/2009/09/625310.html (Last accessed November 18, 2016) 4 Figure 1: Bluefin Tuna Spawning Ground in the Sea of Japan Note: Dark blue area shows the Bluefin tuna spawning ground. Spawning ground information was obtained from the Fisheries Agency of Japan (FAJ, Fishery Agency of Japan, 2014) In recent years, media reports have surfaced of a conflict between the purse seine fishery and pole-and-line fishermen. Relatively smaller scale pole-and-line fishermen argue that an increased harvest of Bluefin tuna by purse seine vessels have caused a stock decline, which has led to the worsening of their catch (?). While there is no scientific argument to support this claim, it has put the problem of Bluefin tuna stock decline on the public radar, resulting in an increase in pressure to tighten the regulation of Bluefin tuna fisheries. 3 Data This study uses auction data from the Tsukiji Central Wholesale Market in conjunction with purse seine landings data. A complete set of variables is listed in Table A1 in the appendix. Tsukiji data is an under-explored data set, and is only used by a few studies (Ex. Wakamatsu and Miyata (2016)). Existing papers on the analysis of the Japanese seafood 5 market, in general, used either import-export data compiled by the national government (Ex. Asche et al. (2005)) or the landings information of Yaizu Port, which is one of the ports that host distant water vessels (Ex. Bose and McIlgorm (1996)). Tsukiji Market is the largest wholesale seafood market in the world. The market, as measured by annual trade volume, is five times larger than the second largest fish market in the world, the Fulton Fish Market in New York City. The Fulton Fish Market is studied extensively in Graddy (1995), Angrist et al. (2000), and Graddy (2006). Unlike Fulton Fish Market, which uses one-on-one price negotiation between a seller and buyer, prices of fish are determined through auction at the Tsukiji Market. Only fish that did not get sold via an auction are directed to one-on-one negotiation. Tsukiji Central Wholesale Market’s daily auction data is compiled and made available to the public by the Tokyo Metropolitan Government’s Central Wholesale Market Division2 . The data covers daily auction results from 2004 to the present. There are a few weaknesses to this data set. First, the sales volume is aggregated for each species, and it does not allow us to distinguish between the domestic catch and imports. Second, about 20% of the observations have missing price information. It is worthwhile noting that the daily mode price, instead of a mean price, is reported. To overcome the shortcomings of the daily data, monthly summary information from Tsukiji Market is used3 . This data set includes the total sales volume and the average price of domestic and imported fish auctioned at Tsukiji Market from 2002 to 2015. The auctioned volume from the daily data, which includes both the domestic catch and imports, is adjusted by multiplying the daily total auctioned volume by the ratio of the domestic catch in the total auctioned volume as calculated from the monthly summary data. The analysis uses this adjusted daily auctioned volume data instead of the aggregated data to approximate domestic catch. In conjunction with the Tsukiji Market data, purse seine landings information is obtained from the Sakaiminato City Government. For the years from 2011 to 2015, the date of landing and the landed volume is available for every single landing of Bluefin tuna by purse seine vessels. To compensate for the short time series of the purse seine landings information, yearly summary data is also obtained from the Sakaiminato City Government. This time series covers the years from 1982 to 2015. This yearly summary data includes the average ex-vessel price, the annual total landed volume, and the number of landings. Table 1 shows the descriptive statistics of the key variables. The data from the first day 2 Daily auction data (in Japanese) is available from http:www.shijou-nippo.metro.tokyo.jp/SN/SN_ Sui_Nengetu.html 3 Monthly auction data (in Japanese) is available from http://www.shijou-tokei.metro.tokyo.jp/ 6 Table 1: Tsukiji Whole Sale Market Data Descriptive Statistics (2004 - 2014 during Sakaiminato Season) Mean Daily Max Price in JPY (USD) Log (Daily Max Price in JPY) Daily Mode Price in JPY (USD) Log (Daily Mode Price in JPY) Daily Sales Vol. (kg) Log (Daily Sales Vol.) Std. Dev. 7,683 (75.34) 8.875 2,469 (23.66) 7.703 16,368 9.522 3,040 0.378 1,373 0.446 9,997 0.618 Max Min 25,000 (239.54) 10.127 13,333 (127.75) 9.498 76,678 11.247 2,700 (25.87) 7.901 608 (5.83) 6.410 1,958 7.580 Figure 2: Purse Seine Bluefin Tuna Landings at Sakaiminato (in metric ton) of the sales at the Tsukiji Market is dropped, as the prices on that day are significantly higher than the other days. The first auction of the year is called a celebratory market, and many auctioneers bid much higher than usual. In the year 2013, the celebratory market price went up to 700,000 JPY (approximately 7,000 USD) per kilogram. Figure 2 shows the yearly landings by purse seine vessels. The landed volume of Bluefin tuna by purse seine vessels increased dramatically in 2004 and 2005, and though it has since decreased, it remains elevated above previous levels . While the cause of this is not clear, this increase in landings coincides with an increase in fishing effort. In a typical year, purse seine fishing of Bluefin tuna starts in late May and lasts until the end of July. In each year, 8 to 12 purse seine vessels participate in the fishery. Each vessel spends approximately two weeks at sea. A vessel comes back to the port once it has harvested 7 Figure 3: Purse-Seine Bluefin Tuna Landings at Sakaiminato (Number of Times) Figure 4: Purse Seine Bluefin Tuna Catch per Fishing Trip (Metric Ton) 8 a reasonable amount of Bluefin tuna, though this amount varies depending on various factors. As Figure 3 shows, the number of times that the vessels return to port to land their catch increased by 2.5 times in 2004. Given this shift in the purse seine vessels’ effort level, we only include the data after the year 2004 in the analysis that use variations in the purse seine catch as instruments. Also, because the purse seine Bluefin fishery adopted new voluntary measures to limit the catch in 2015, we exclude data from the year 2015 in the analysis. 4 Estimation Framework In the estimation of the price elasticity of demand, the naı̈ve approach would be to regress quantity against price using the ordinary least square method. However, there are two problems associated with this approach because the observed quantities and prices are at the intersection of supply and demand curves. Firstly, because there are more coefficients than the number of equations, the coefficients will not be identified4 . Second, the ordinary least square estimator is biased and inconsistent due to the endogeneity problem. To overcome these problems, an instrumental variables approach can be applied. The instrumental variables approach works as follows. First, by denoting the log transformed prices and quantities as q and p, with superscript S and D to denote supply and demand respectively, along with the error term u and v, the estimation equations for supply and demand can be expressed as Supply equation: qtS = β0 + β1 ptS + γzt + ut Demand equation: qtD = α0 + α1 ptD + vt . In the supply equation, zt represents the exogenous supply shock. The prices and quantities that we observe in the data are at the intersections of supply and demand curves. Therefore, we have q∗ = qS = qD and p∗ = pS = pD , where ∗ denotes equilibrium. By solving the two equations, we have pt∗ = β 0 − α0 γ 1 + zt + (u − vt ) ≡ δ0 + δ1 zt + t α1 − β1 α1 − β1 α1 − β 1 t The right hand side of the equation is the reduced form estimation model. However, 4 The discussion on the identification of the constant price elasticity of demand is detailed in the Appendix A. 9 this model lacks an economic intuition. A better estimation approach would be to use the two-stage least square method. The reduced form estimation model serves as a first-stage estimation model for the two-stage least square estimation. The first stage estimation model can be expressed as: p̂t = δ0 + δ1 zt + t . The second stage estimation model uses the estimated price pˆt so that the second stage estimation model can be expressed as qt = α0 + α1 p̂t + σ X + errort . The basic idea used in this study is similar to the one used in Angrist et al. (2000), which used weather conditions as exogenous supply shocks. A key qualification for any instrument used to estimate demand elasticity is that it must be an exogenous supply shock that shifts the supply curve without affecting the demand curve. In this study, we use the intra- and inter-annual variations in landings in the purse seine Bluefin tuna fishery as instruments. There are four important characteristics that make the instruments valid. First, purse seine landings account for a large portion of the domestic Bluefin tuna catch. Second, the purse seine fishing season only lasts for approximately 70 days. Third, a landing by a purse seine vessel is far larger than a landing by any other gear. Fourth, landings by purse seine vessels do not take place every day of the season. In the years from 2011 to 2014, there were only 22 landings, on average, per season. These characteristics indicate that each landing by a purse seine vessel shift the supply of fresh Bluefin tuna in the Japanese market. We estimate with two variables as instruments: 1) the landed volume by a purse seine vessel on the previous day, and 2) a dummy variable indicating whether or not there was a landing by a purse seine vessel on the previous day. In addition, inter-annual variations in catch by purse seine vessels serve as a quasiexperiment for supply shocks. From Figure 2, even after the increase in the trips made by purse seine vessels after the year 2004, the landed volume is low in some years. In particular, the landings in years 2009, 2010, and 2012 fall below the 1st quintile of the landings after 2004. In these years, the catch per fishing trip is also low (Figure 4). Because such drops are most likely due to unknown oceanographic conditions, as previously discussed, we could clearly rule out the possibility of price influencing this variable. From this, we estimate price elasticity of demand by using 1) a dummy variable indicating low catch years, and 2) average annual catch per fishing trip as instruments. We can also use fluctuations in daily auctioned volume at the Tsukiji Market as in10 Figure 5: Tsukiji Daily Auctioned Volume (in Kilogram) Note: The colored lines show approximated time trends with spline knots equal to 4 (blue), 9 (red), and 132 (green). The gray shaded area is the 95% confidence interval. struments. Because of the limited storability for Bluefin tuna, and the fluctuations in the auctioned volume at Tsukiji Market are most likely due to the fluctuations in landings. From this, we can calculate deviations from the time trend as a proxy for supply shocks. To do this, we first calculate the time trend by using the cubic spline method5 . Figure 5 shows the daily auctioned volume of fresh Bluefin tuna at Tsukiji Market. The graph shows great variations in daily auctioned volume. The colored lines are time series trends approximated by using natural cubic splines with spline knots equal to 4 (blue), 9 (red), and 132 (green). The gray shaded area is the 95% confidence interval of the time trend approximation. The fluctuations are calculated by taking the difference between the observed data and the time trend. One of the unique aspects of Bluefin tuna fishing in Japan is that all coastal and offshore Bluefin tuna fisheries are seasonal. This causes a monthly fluctuation in Bluefin catch. This can be observed clearly in the monthly summary data (Figure 6). Again, the colored lines show approximated time trends with spline knots equal to 4 (blue), 5 (red), and 13 (green). The gray shaded area is the 95% confidence interval of the time trend approximation. By following the same steps used to calculate the daily fluctuations, we can also calculate the monthly fluctuations to be used as instruments. In total, six instruments are used to estimate the own-price elasticity of demand. Two stage least square methods are used to estimate the value by using the six instruments. 5 Pollock (2007) discusses that the spline method is an appropriate strategy to model a local characteristics of time trends without sacrificing global trend. 11 Figure 6: Tsukiji Monthly Auctioned Volume (in Metric Ton) Note: The colored lines show approximated time trends with spline knots equal to 4 (blue), 5 (red), and 13 (green). The gray shaded area is the 95% confidence interval. For the instrumental variables to yield consistent estimates and behave better than the ordinary least square estimators, instruments need to be correlated with the endogenous regressor (Murray, 2006). To verify our choice of instruments, two tests are performed. The weak instrument test performs an F-test on the first-stage regression and checks the correlation between the instrument and the endogenous regressor. The Wu-Hausman test checks whether the instrumental variables yield more consistent results than the ordinary least square estimator. 5 5.1 Results Main Results Table 2 shows the reduced form, or first-stage, estimate and the estimated price elasticity of demand with four instruments. For all the specifications, we limit our analysis to the purse seine fishing season. The season is determined by the first and the last day of the landings. Two of the instruments are created from the characteristic that purse seine landings do not take place everyday. Column (4) uses the landed volume by purse seine vessels on a previous day as an instrument. Column (5) uses a dummy variable that indicates whether there was a landing by a purse seine vessel on a previous day. The other two instruments are based on the inter-annual variabilities in purse seine 12 landings. We define the years that fell below 1,219 metric tons, those that make up the first quintile of the annual harvest for the years 2004 to 2014, as the low catch years. By this definition, the years 2009, 2010, and 2012 are defined as the low-catch years. Column (7) uses a dummy variable that indicates the low purse seine catch years as instrument. Column (8) uses annual average landed volume per trip as an instrument. Because purse seine vessels may come back without filling to the vessel’s storage capacity when stock conditions are not ideal, this variable would be an appropriate measure for stock conditions. The estimated elasticity of demand ranges between -1.247 to -2.163. The adjusted R square values are low for all models. The adjusted R square value is negative for the estimation (7). However, this is not of a concern because the adjusted R square values do not have any statistical meaning in the estimations using the 2-stage least square method (Wooldridge, 2010). From the reduced form estimation, we find that the coefficient for the variable of the landed volume by a purse seine vessel, or vessels, is very small. There are two possibilities for why this is the case. First, the correlation between this instrument and the endogenous variable is weak. If this is the case, it poses a serious problem, as the weak correlation causes the instrumental variable estimator to behave worse than the ordinary least square estimator (Yogo, 2004). Second, the instrument may not be properly controlled for by other variables. This is less problematic if we can identify and add sufficient control variables. Despite the small coefficient value, the variable passes the F-test that tests for weak instruments. All other instruments pass both the weak instrument and the Wu-Hausman test, suggesting that the instrumental variable estimator is more consistent than the ordinary least square estimator. Table 3 summarizes the results of the same four instrumental variables with varying specifications to control for confounding factors. The estimated elasticity of demand ranges from -1.238 to -2.149. The logged daily maximum auction price is included for the specifications (2), (4), (6) and (8). This variable is included as a proxy for the quality of fish. A day of the year variable is included to check whether the auctioned volume changes as the purse seine season progresses. A day of the week variable is included because previous studies indicate demand differs throughout the week (Ex. Angrist et al. (2000)). All specifications pass the weak instrument test and the Wu-Hausman test. 13 14 171 0.307 0.303 Observations R2 Adjusted R2 Note: ∗ p<0.1; ∗∗ p<0.05; ∗∗∗ p<0.01 NA NA 7.849∗∗∗ (0.036) 171 0.234 0.230 NA NA 7.902∗∗∗ (0.044) −0.465∗∗∗ (0.065) (2) 411 0.022 0.019 NA NA 7.660∗∗∗ (0.026) 0.142∗∗∗ (0.047) (3) 411 0.035 0.032 NA NA −0.013∗∗∗ (0.004) 8.110∗∗∗ (0.108) (4) 171 0.142 0.137 *** *** 19.991∗∗∗ (1.269) 171 0.216 0.211 *** *** 19.121∗∗∗ (1.388) X −1.247∗∗∗ (0.180) −1.360∗∗∗ (0.165) X (6) (5) 411 −0.679 −0.684 *** *** 26.156∗∗∗ (4.736) X −2.163∗∗∗ (0.615) (7) Instrumental Variable 2SLS Reduced Form (1st Stage) OLS −0.00001∗∗∗ (0.00000) Weak Instrument Test Wu-Hausman Test Constant PS ave. land. vol. per trip (IV) PS low catch year (IV) PS prev. day. land. dum.(IV) PS prev. day land. vol. (IV) Price elas. of demand (1) Log(Auctioned Volume) Log(Price) Dependent Variable: Table 2: Reduced Form and IV Estimates (No Covariates) 411 −0.033 −0.035 *** *** 21.671∗∗∗ (2.926) X −1.581∗∗∗ (0.380) (8) 15 Note: ∗ p<0.1; ∗∗ p<0.05; ∗∗∗ p<0.01 Observations R2 Adjusted R2 IV PS land. vol. prev. day IV PS land. prev. day dum. IV Low PS catch yr. IV Annual PS land. per trip Weak Instrument Test Wu-Hausman Test Constant Saturday Friday Thursday Wednesday Tuessday Day of the year Log(Daily max price) Elas. of demand *** *** −0.007∗∗∗ (0.002) −0.311∗∗ (0.128) −0.424∗∗ (0.190) 0.173 (0.128) 0.040 (0.134) 0.003 (0.130) 20.325∗∗∗ (1.143) X *** *** 171 0.364 0.332 −1.255∗∗∗ (0.163) 0.132 (0.106) −0.007∗∗∗ (0.002) −0.305∗∗ (0.129) −0.410∗∗ (0.192) 0.156 (0.130) 0.053 (0.135) 0.016 (0.132) 19.380∗∗∗ (1.360) X −1.238∗∗∗ (0.161) 171 0.369 0.342 (2) (1) 171 0.321 0.292 *** *** X −0.006∗∗∗ (0.002) −0.310∗∗ (0.132) −0.395∗∗ (0.198) 0.175 (0.133) 0.053 (0.139) 0.009 (0.135) 20.764∗∗∗ (1.217) −1.304∗∗∗ (0.172) (3) 171 0.308 0.273 *** *** X −1.330∗∗∗ (0.175) 0.138 (0.111) −0.007∗∗∗ (0.002) −0.303∗∗ (0.134) −0.377∗ (0.201) 0.158 (0.135) 0.067 (0.141) 0.024 (0.138) 19.840∗∗∗ (1.442) (4) 411 −0.614 −0.642 *** *** X 0.0003 (0.003) −0.274∗∗ (0.136) −0.446∗∗ (0.210) −0.023 (0.137) −0.077 (0.140) −0.127 (0.136) 26.128∗∗∗ (4.395) −2.149∗∗∗ (0.635) (5) Log(Auctioned volume) Dependent Variable: Table 3: IV Estimates (with Covariates) 411 −0.391 −0.418 *** *** X −1.995∗∗∗ (0.544) 0.228∗ (0.120) −0.001 (0.003) −0.264∗∗ (0.126) −0.433∗∗ (0.197) −0.021 (0.127) −0.054 (0.132) −0.098 (0.128) 23.188∗∗∗ (3.351) (6) 411 0.081 0.065 X *** *** −0.002 (0.002) −0.270∗∗∗ (0.102) −0.569∗∗∗ (0.146) −0.028 (0.103) −0.115 (0.104) −0.146 (0.102) 21.863∗∗∗ (2.595) −1.530∗∗∗ (0.374) (7) 411 0.142 0.125 X *** *** −1.464∗∗∗ (0.345) 0.166∗ (0.090) −0.003 (0.002) −0.263∗∗∗ (0.099) −0.550∗∗∗ (0.145) −0.026 (0.100) −0.095 (0.102) −0.123 (0.100) 20.042∗∗∗ (2.167) (8) 5.2 Fluctuations as Supply Shock Table 4 summarizes reduced form and instrumental variable estimations with daily fluctuations in auctioned volume as instruments. These specifications include observations from the years 2004 to 2014 for all days that the data is available6 . The coefficients for supply shocks (i.e. daily fluctuations) in the reduced form estimation show significant but small values. However, these instruments pass the weak instrument test. Using fluctuations as supply shocks yields an estimated price elasticity of demand ranging from -2.603 to -4.793. These estimates are slightly higher in absolute value than the specifications using the purse seine landings as instruments. Estimation results from the daily fluctuations in auctioned volume are presented in Table 5. The results with different covariates and time trend estimates are presented. The estimated elasticity of demand values are more elastic using these fluctuations than when using purse seine landings as fluctuations. The values range from -2.343 to -3.591. The estimates from the specifications that only consider the purse seine season have lower absolute values. The estimates from the specifications that consider all days, with a dummy variable that indicates the purse seine seasons are negative and significant. Given these observations, the results hint that the elasticity of demand may be slightly less elastic during the purse seine season. A typical purse seine season normally lasts from the late May to the end of July. The more inelastic demand estimate during purse seine seasons may be explained by the fact that the overall seafood supply is low in the summer months. Low overall seafood supply limits availability of alternatives, which makes the market more dependent on Bluefin tuna. Indeed, an interview with the local fishery research institute suggests that purse seine Bluefin tuna fishery markets to fill seasonal void in the seafood market7 . 6 In the data set, the price information is missing for about 20% of the days. of the Tottori Prefecture Fishery Division Sakaiminato Office by the author (November 2015) 7 Interview 16 17 Note: ∗ p<0.1; ∗∗ p<0.05; ∗∗∗ p<0.01 3 2,983 0.164 0.164 No. of spline knots Observations R2 Adjusted R2 9 2,983 0.163 0.163 NA NA 8.158∗∗∗ (0.006) 8.159∗∗∗ (0.006) NA NA −0.00003∗∗∗ (0.00000) −0.00003∗∗∗ (0.00000) (2) 132 2,983 0.027 0.027 NA NA 8.163∗∗∗ (0.009) −0.00002∗∗∗ (0.00000) (3) 3 2,344 −3.191 −3.193 *** *** 9 2,344 −3.145 −3.147 *** *** 30.248∗∗∗ (1.045) −2.603∗∗∗ (0.128) −2.619∗∗∗ (0.128) 30.376∗∗∗ (1.048) (5) (4) 132 2,344 −12.739 −12.745 *** *** 48.123∗∗∗ (4.657) −4.793∗∗∗ (0.570) (6) Instrumental Variable 2SLS Reduced Form (1st Stage) OLS Weak Instrument Test Wu-Hausman Test Constant Supply shock Elas. of demand (1) Log(Auctioned Volume) Log(Price) Dependent Variable: Table 4: Daily Fluctuations as Instruments (No Covariates) 18 4 No 2,344 −2.127 −2.150 No. of spline knots Only incl. PS season Observations R2 Adjusted R2 Note: ∗ p<0.1; ∗∗ p<0.05; ∗∗∗ p<0.01 X X *** *** 22.753∗∗∗ (0.788) 4 Yes 411 −1.031 −1.119 X X *** *** 25.925∗∗∗ (1.777) −2.484∗∗∗ (0.223) 0.300∗∗ (0.136) −2.498∗∗∗ (0.123) 0.718∗∗∗ (0.076) Day of the week FE Year FE Weak Instrument Test Hausman-Wu Constant PS season dummy Log(Daily max P) Elas. of demand (2) (1) 4 No 2,344 −4.892 −4.938 X X *** *** −3.591∗∗∗ (0.271) 0.836∗∗∗ (0.111) −1.222∗∗∗ (0.157) 30.747∗∗∗ (1.752) (3) 9 No 2,344 −2.001 −2.022 X X *** *** 22.507∗∗∗ (0.766) −2.444∗∗∗ (0.119) 0.697∗∗∗ (0.074) (4) 9 Yes 411 −1.020 −1.108 X X *** *** 25.887∗∗∗ (1.770) −2.478∗∗∗ (0.222) 0.298∗∗ (0.136) (5) 9 No 2,344 −4.531 −4.574 X X *** *** −3.477∗∗∗ (0.256) 0.806∗∗∗ (0.106) −1.166∗∗∗ (0.150) 30.085∗∗∗ (1.666) (6) Log(Auctioned Volume) Dependent Variable: Table 5: Daily Fluctuations as Instruments (with Covariates) 132 No 2,344 −4.658 −4.700 X X *** *** 26.777∗∗∗ (1.647) −3.371∗∗∗ (0.320) 1.062∗∗∗ (0.149) (7) 132 Yes 411 −0.808 −0.886 X X *** *** 25.090∗∗∗ (1.726) −2.343∗∗∗ (0.221) 0.272∗∗ (0.129) (8) 132 No 2,344 −4.690 −4.735 X X *** *** −3.527∗∗∗ (0.343) 0.819∗∗∗ (0.123) −1.191∗∗∗ (0.188) 30.379∗∗∗ (2.129) (9) Estimation results from monthly fluctuations in auctioned volume are presented in Table 6. The estimated elasticity of demand values form a narrow range from -1.887 to -1.938. The results of the weak instrument test show that the instruments pass for all specifications. The results of the Wu-Hausman test indicate that the instrumental variables are consistent, while the ordinary least square estimator would be inconsistent, thereby supporting the use of monthly fluctuations as instruments. The estimated values of elasticity of demand are close to the estimated values from the specifications using purse seine landings as instruments. 19 20 Note: ∗ p<0.1; ∗∗ p<0.05; ∗∗∗ p<0.01 NA 156 0.515 0.512 No. of spline knots Observations R2 Adjusted R2 NA 156 0.517 0.514 NA NA 3,537.282∗∗∗ (47.431) 3,536.606∗∗∗ (47.508) NA NA −0.006∗∗∗ (0.0005) −0.006∗∗∗ (0.0005) (2) NA 156 0.517 0.514 NA NA 3,535.190∗∗∗ (47.415) −0.006∗∗∗ (0.001) (3) 4 156 0.231 0.226 *** *** 5 156 0.231 0.226 *** *** 27.867∗∗∗ (1.454) −1.938∗∗∗ (0.179) −1.936∗∗∗ (0.179) 27.854∗∗∗ (1.455) (5) (4) 13 156 0.251 0.246 *** *** 27.430∗∗∗ (1.430) −1.884∗∗∗ (0.176) (6) 4 156 0.604 0.567 X *** *** 26.865∗∗∗ (1.081) −1.887∗∗∗ (0.132) (7) Instrumental Variable 2SLS Reduced Form (1st Stage) OLS Year FE Weak Instrument Test Wu-Hausman Test Constant Supply shock Log(Price) (1) Log(Auctioned Volume) Log(Price) Dependent Variable: Table 6: Monthly Fluctuations as Instruments 5 156 0.604 0.567 X *** *** 26.866∗∗∗ (1.081) −1.888∗∗∗ (0.132) (8) 13 156 0.604 0.567 X *** *** 26.865∗∗∗ (1.081) −1.887∗∗∗ (0.132) (9) 6 Conclusion and Discussion This study estimated the elasticity of demand for fresh Bluefin tuna in Japan by using the instrumental variables approach. Because Japan is the largest consumer of Pacific Bluefin tuna, understanding the characteristics of Japanese demand for Bluefin helps to predict the overall market consequences of tighter conservation measures. In the estimates that used variations in purse seine landings as supply shock instruments, the estimated constant own-price elasticity of demand ranged from -1.247 to -2.163, depending on the specifications. In the estimates that used the daily fluctuations in the auctioned volume at the Tsukiji Market as supply shock instruments, the estimated constant own-price elasticity of demand ranged from -2.444 to -4.739, when we did not control for the purse seine season. The results from the specifications that controlled for the purse seine season indicated slightly less elastic demand during the purse seine season. In the estimates that used the monthly fluctuations in the auctioned volume at the Tsukiji Market as supply shock instruments, the estimated constant own-price elasticity of demand ranged from -1.887 to -1.938. To the author’s knowledge, there is currently no estimate of the elasticity of global demand for Bluefin tuna. This study could reasonably serve as an approximation of the elasticity of global demand as the Japanese market accounts for 80% of the fresh Bluefin tuna traded in the world. One previous study found the retail price elasticity of demand for tuna to be inelastic (Wessells and Wilen, 1994). Yet another study by Sun et al. (2015) found unit elastic demand for Skipjack and Yellowfin tuna in the canned tuna market. Because Bluefin tuna has the highest price of any tuna, and can be perceived as a luxury food item, our findings of an elastic demand fit with intuition. An elastic demand may be a problem for Bluefin tuna fisheries, as they may not be able to attain enough price increases if Bluefin supply is reduced to cover the loss from the reduced catch. This may pose a challenge to the industry if catch limits are imposed. The results also suggest the possibility that demand for Bluefin tuna in the summer may be less elastic than in other seasons. If there is indeed a variability in demand elasticity across seasons, with a more inelastic summer demand, a catch limit imposed on purse seine fisheries may be less damaging overall. Crona et al. (2015) warns that seafood prices are not working as a signal for ecosystem health. The conclusion that Bluefin tuna demand is elastic may be interpreted as evidence that Bluefin tuna prices are not working as a signal for the species’ population health. To fully address this, future studies should look at the substitutability among different 21 species of fish. As we see in the Bluefin tuna fisheries in Japan, many coastal fisheries often target different species at different times throughout the year. Those seasonal target choices may be studied further to understand the supply elasticities for different seafood products. Furthermore, in the presence of seasonal fisheries, a species-based regulation may impacts different fisheries differently, or uniformly, if demand elasticity differ across seasons. To address this, future studies should further investigate seasonality in the seafood market. 22 References Angrist, J., Graddy, K., and Imbens, G. W. (2000). The Inerpretation of Instrumental Variables Estimators in Simultaneous Models with an Application to the Demand for Fish. The Review of Economic Studies, 67:499 – 527. Angrist, J. and Krueger, A. B. (2001). Instrumental variables and the search for identification: From supply and demand to natural experiments. Technical report, National Bureau of Economic Research. Asche, F., Bjørndal, T., and Gordon, D. V. (2007). Studies in the demand structure for fish and seafood products. In Handbook of Operations Research in Natural Resources, pages 295–314. Springer. Asche, F., Guttormsen, A. G., Sebulonsen, T., and Sissener, E. H. (2005). Competition between farmed and wild salmon: the Japanese salmon market. Agricultural Economics, 33:333 – 340. Barten, A. P. (1968). Estimating demand equations. Econometrica: Journal of the Econometric Society, pages 213–251. Bose, S. and McIlgorm, A. (1996). Substitutability among species in the Japanese tuna market: a cointegration analysis. Marine Resource Economics, pages 143–155. Crona, B. I., Daw, T. M., Swartz, W., Norström, A. V., Nyström, M., Thyresson, M., Folke, C., Hentati-Sundberg, J., Österblom, H., Deutsch, L., and others (2015). Masked, diluted and drowned out: how global seafood trade weakens signals from marine ecosystems. Fish and Fisheries. Deaton, A. and Muellbauer, J. (1980). An almost ideal demand system. The American economic review, 70(3):312–326. Eales, J., Durham, C., and Wessells, C. R. (1997). Generalized models of Japanese demand for fish. American Journal of Agricultural Economics, 79(4):1153–1163. FAJ, Fishery Agency of Japan (2014). Taiheiyou kuromaguro no kanri kyoka no torikumi jokyo to kongo no taiou ni tsuite. Fishery Agency of Japan Website. Graddy, K. (1995). Testing for imperfect competition at the Fulton fish market. The RAND Journal of Economics, pages 75–92. 23 Graddy, K. (2006). Markets: The Fulton Fish Market. Journal of Economic Perspectives, 20(2):207–220. IUCN, International Union for Conservation of Nature (2014). Thunnus orientalis: Collette, B., Fox, W., Juan Jorda, M., Nelson, R., Pollard, D., Suzuki, N. & Teo, S.: The IUCN Red List of Threatened Species 2014: e.T170341a65166749. Technical report. Makino, M. and Matsuda, H. (2005). Co-management in Japanese coastal fisheries: institutional features and transaction costs. Marine Policy, 29(5):441–450. McCurry, J. (2015). Bluefin; japan’s favorite big fish, but for how long. The Guardian. Murray, M. P. (2006). Avoiding invalid instruments and coping with weak instruments. The journal of economic perspectives, 20(4):111–132. OPRT, Organization for the Promotion of Responsible Tuna Fisheries (2004). Sekai no maguro gyokaku-ryo to nihon no sashimi maguro shijo. OPRT Website. Pollock, D. S. G. (2007). Trends in economic time series. Lecture Note in Time Series and Forecasting. Roberts, M. J. and Schlenker, W. (2013). Identifying Supply and Demand Elasticities of Agricultural Commodities: Implications for the US Ethanol Mandate. The American Economic Review, 103(6):2265–2295. Sun, C.-H. J., Chiang, F.-S., Guillotreau, P., Squires, D., Webster, D. G., and Owens, M. (2015). Fewer Fish for Higher Profits? Price Response and Economic Incentives in Global Tuna Fisheries Management. Environmental and Resource Economics, pages 1–16. Theil, H. (1965). The information approach to demand analysis. Econometrica: Journal of the Econometric Society, pages 67–87. Tottori Prefecture Fisheries Research Institute (1995). Sakaiminato no suisan. Tottori Prefecture Fisheries Research Institute Report. Wakamatsu, H. and Miyata, T. (2016). Do Radioactive Spills from the Fukushima Disaster Have Any Influence on the Japanese Seafood Market? Marine Resource Economics, 31(1):27–45. Wessells, C. R. and Wilen, J. E. (1994). Seasonal patterns and regional preferences in Japanese household demand for seafood. Canadian Journal of Agricultural Economics/Revue canadienne d’agroeconomie, 42(1):87–103. 24 Wooldridge, J. M. (2010). Econometric Analysis of Cross Section and Panel Data. MIT Press, 2 edition. Yogo, M. (2004). Estimating the Elasticity of Intertemporal Substitution When Instruments Are Weak. Review of Economics and Statistics, 86(3):797–810. 25 Appendix A: Identification of the Elasticity of Demand First, by denoting the log transformed prices and quantities as q and p with superscript S and D to denote supply and demand respectively, the estimation equations for the the supply and demand can be expressed along with the error term u and v as Supply equation: qtS = β0 + β1 ptS + γzt + ut Demand equation: qtD = α0 + α1 ptD + vt In the supply equation, zt represents exogenous supply shocks. The prices and quantities that we observe in the data are the intersections of supply and demand curve. Therefore, we have q∗ = qS = qD and p∗ = pS = pD . By solving the two equations, have pt∗ = and qt∗ = β0 − α0 γ 1 + zt + (u − vt ) ≡ δ0 + δ1 zt + t , α1 − β1 α1 − β1 α1 − β1 t α1 γ (β0 − α0 )α0 α1 + zt + (u − vt ) ≡ θ0 + θ1 zt + ηt . α1 − β 1 α1 − β1 α1 − β1 t From the exogenous supply shock zt , we are able to identify α0 from θ0 /δ0 and α1 from θ1 /δ1 . Hence, by using an exogenous supply shock, we are able to identify the elasticity of demand. 26 Appendix B: Supplementary Figures and The List of Variables Figure A1: Purse Seine Landings vs Tsukiji Adjusted Auctioned Volume 27 Figure A2: Purse Seine Landing on Previous Day vs Tsukiji Adjusted Auctioned Volume 28 Figure A3: Tsukiji Market Auctioned Price and Volume 29 30 JPY/kilogram JPY/kilogram Mid price** Low price*** *Tokyo Metropolitan Government defines high price as the highest price of the day **Tokyo Metropolitan Government defines mid price as mode price of the day ***Tokyo Metropolitan Government defines low price as mode of the prices below the mid price of the day Tsukiji Market Times Metric ton Number of fish 1000 JPY Date Date Metric ton Kilogram JPY/kilogram Number of landings Landed volume Number of landed fish Landed value First day of landing Last day of landing Landed volume Sales volume High price* Year Sakaiminato total Year Sakaiminato total Year Sakaiminato total Year Sakaiminato total Year Sakaiminato total Year Sakaiminato total Day Vessel Day Aggregate Day Major domestic port, other domestic port total, overseas total Day Major domestic port, other domestic port total, overseas total Day Major domestic port, other domestic port total, overseas total Sakaiminato Units Variable Name Dataset Table A1: List of Variables 1982 - 2015 1982 - 2015 1982 - 2015 1982 - 2015 1982 - 2015 2000, 2002 - 2015 2011 - 2015 2004 - 2015 2005 - 2015, include missing data 2006 - 2015, include missing data 2007 - 2015, include missing data Time Period
© Copyright 2026 Paperzz