UBC Community, Partners, and Alumni Publications

Economic valuation of critical habitat closures Berman, Matthew; Gregr, Edward; Ishimura, Gakushi; Coatta, Ryan; Flinn, Rowenna; Sumaila, U. Rashid; Trites, Andrew 2008

You don't seem to have a PDF reader installed, try download the pdf

Item Metadata

Download

Media
FCRR_2008_16-8.pdf [ 959.2kB ]
[if-you-see-this-DO-NOT-CLICK]
Metadata
JSON: 1.0074746.json
JSON-LD: 1.0074746+ld.json
RDF/XML (Pretty): 1.0074746.xml
RDF/JSON: 1.0074746+rdf.json
Turtle: 1.0074746+rdf-turtle.txt
N-Triples: 1.0074746+rdf-ntriples.txt
Original Record: 1.0074746 +original-record.json
Full Text
1.0074746.txt
Citation
1.0074746.ris

Full Text

ISSN 1198-6727  Fisheries Centre Research Reports  2008 Volume 16 Number 8  Economic Valuation of Critical Habitat Closures  Fisheries Centre, University of British Columbia, Canada  Economic Valuation of Critical Habitat Closures  By Matthew Berman1 Edward Gregr2 Gakushi Ishimura2 Ryan Coatta2 Rowenna Flinn2 U. Rashid Sumaila2 and Andrew Trites2 1. Institute of Social and Economic Research, University of Alaska Anchorage, Anchorage AK, USA 2. Fisheries Centre, The University of British Columbia, Vancouver, BC, Canada  Fisheries Centre Research Reports 16(8) 102 pages © published 2008 by The Fisheries Centre, University of British Columbia 2202 Main Mall Vancouver, B.C., Canada, V6T 1Z4  ISSN 1198-6727  Fisheries Centre Research Reports 16(8) 2008 ECONOMIC VALUATION OF CRITICAL HABITAT CLOSURES  By  Matthew Berman, Edward J. Gregr, Gakushi Ishimura, Ryan Coatta, Rowenna Flinn, U. Rashid Sumaila, and Andrew Trites  CONTENTS Page DIRECTOR’S FOREWORD ...................................................................................................................................... 1 ABSTRACT ........................................................................................................................................................... 2 INTRODUCTION ................................................................................................................................................... 3 METHODS ........................................................................................................................................................... 5 RESULTS ........................................................................................................................................................... 10 DISCUSSION ...................................................................................................................................................... 33 CONCLUSIONS ................................................................................................................................................... 34 ACKNOWLEDGEMENTS ....................................................................................................................................... 34 REFERENCES ..................................................................................................................................................... 35 APPENDICES ...................................................................................................................................................... 37 Appendix A. Equations for Spatial Distribution of Catch Per Unit of Effort (CPUE), Estimated from the Summer 2001 NMFS Gulf of Alaska Bottom Trawl Survey .............................. 37 1. Equations with Wind -- only shown if absolute value of t on wind > 1.........................................38 2. Equations without wind ................................................................................................................ 45 Appendix B. Equations for Spatial Distribution of Catch Per Unit of Effort (CPUE) for Summer and Winter, Bering Sea/Aleutian Islands and Gulf of Alaska, Estimated from 2001 NMFS Bottom Trawl Fisheries Observer Data .......................................................................... 58 1. Winter bottom trawl.......................................................................................................................58 2. Summer bottom trawl selection equation (IMR7)........................................................................ 81  A Research Report from the Fisheries Centre at UBC 102 pages © Fisheries Centre, University of British Columbia, 2008  FISHERIES CENTRE RESEARCH REPORTS ARE ABSTRACTED IN THE FAO AQUATIC SCIENCES AND FISHERIES ABSTRACTS (ASFA)  ISSN 1198-6727  Economic Valuation Of Critical Habitat Closures, Berman et al.  1  DIRECTOR’S FOREWORD Humans have developed fisheries spanning entire oceans, and have the capability to overexploit the resources in any region within a very short time, as attested by numerous now-defunct fisheries. Therefore, a consensus is slowly emerging that management of fisheries, rather than focusing only on the amount of fishing effort deployed, also needs to be structured in space, with different ocean areas being targeted differently, and/or at different times, depending on the resources and habitat that they provide. In fact, ocean zoning is emerging as a major element of Ecosystem-Based (Fisheries) Management, because ecosystems are spatial entities. Ecosystem-Based (Fisheries) Management implies, among other things, redirecting fishing effort away from previously fished areas to protect animals or habitats whose continued existence is considered crucial. For the eastern North Pacific ocean, the rookeries and haulouts of Steller sea lions and the feeding areas surrounding them provide a clear example of areas that need protection. The cost to the fisheries of closing such areas can be evaluated and balanced against the risk of damage to natural resources in the area. This cost will be some fraction of the value of the catch that could be made in the areas to be closed, its value depending on the extent to which substitute areas are available to the fishery. This report presents a spatial model of fleet operations through which such costs can be evaluated; as such, it is of interest to anyone interested in spatial management and marine protected areas. Some colleagues claim not to know what Ecosystem-Based (Fisheries) Management means, or even that it does not mean anything concrete. This report shows what Ecosystem-Based (Fisheries) Management can mean, and it poses questions and proposes an approach for answering them that would not have seen the light of day when single-species approaches reigned. So, gradually, we are getting there.  Daniel Pauly Director, UBC Fisheries Centre  2  Economic Valuation Of Critical Habitat Closures, Berman et al.  ABSTRACT We developed methods to estimate the spatial variation in economic values of ocean fisheries, and applied the methods to estimate the cost of closing groundfish fisheries in Steller sea lion Critical Habitat in the Bering Sea and Gulf of Alaska. The research addressed two related goals: (1) explicitly linking spatial variability of fisheries biomass and profitability over time to environmental variables; and (2) developing estimates of opportunity costs of time and area closures to the fishing industry at scales relevant to management. The approach involved two stages of statistical analyses. First, environmental conditions measured at 3 km and 9 km spatial scales and two-week and one-month intervals were used to predict fish biomass and fisheries catch per unit of effort (CPUE). Environmental variables included bathymetry, remotely sensed physical and biological observations, and output from a physical oceanographic circulation model. Second, we used predicted CPUE and spatial regulatory and cost factors to explain the spatial distribution of fishing effort over time. Our results suggested that 2001 Critical Habitat closures cost the North Pacific groundfish trawl fisheries 5-40 percent of their total potential net earnings. The improved methods for estimating opportunity costs of fisheries closures we present have direct applications to evaluating boundary changes to marine protected areas and other spatial management decisions. Limitations include the extensive data requirements and the need to bootstrap confidence intervals. If further research demonstrates the robustness and stability of the estimated relationships over time, the methods could project spatial fishery effects of climate variability and change, leading to dynamic spatial models linking fisheries with ecosystems.  Economic Valuation Of Critical Habitat Closures, Berman et al.  3  INTRODUCTION Resource managers are increasingly requested to make decisions to restrict commercial fishing for the benefit of protected species, with uncertainty about the value of the reserved habitat to the fishing industry, and to the species at risk. Critical Habitat designations for Steller sea lions (SSLs; Eumetopias jubatus) since 2000 in the Gulf of Alaska and Bering Sea have been especially disruptive to fisheries for pollock (Theragra chalcogramma), Atka mackerel (Pleurogrammus monopterygius), and other groundfish (National Marine Fisheries Service 2001a). However, claims of high annual losses by fisheries organizations cannot be independently evaluated due to the absence of a scientifically defensible method to estimate the cost of the closures around critical habitat areas. The main official study documenting the economic impact of SSL critical habitat designation (National Marine Fisheries Service 2001a) contains only qualitative analyses of the closures on industry profits. Pending proposals to close additional areas to fishing in the North Pacific to protect ‘essential fish habitat’ or ‘habitat areas of particular concern’, and possible future closures to protect other marine species could further reduce the area available for fishing. The controversy surrounding these actions suggests that there is an urgent need to develop objective methods to quantify their cost. Quantitative economic analyses of North Pacific habitat closures have largely been limited to describing what has become known as ‘revenue at risk’ (National Marine Fisheries Service 2001a; Tetra Tech 2004; North Pacific Fishery Management Council 2004). Revenue at risk represents an estimate of the ex-vessel gross revenue that could reasonably be expected to derive from fishing in the area proposed to be closed, based on historic catches when the area was open to fishing. This is a completely inadequate measure of the losses that the industry – and society – would endure from such closures. Under fisheries regulated by Total Allowable Catch (TAC), fishing effort generally moves from closed areas to areas that remain open. Total catch and gross revenue will remain the same as before unless the restrictions are so severe that some TAC remains uncaught, an unlikely outcome for overcapitalized fisheries like those of the North Pacific. True ex-vessel gross revenue losses are probably close to zero in most cases. Although most habitat closures are unlikely to substantially affect total catches, market value, and gross revenues, the expansion of time and area closures on the fishery nevertheless imposes real costs on the industry. Such costs may include higher travel costs to reach open areas, higher operating costs from lower catch rates and interrupted trawls, search costs and costs of learning how to fish profitably in new areas. These costs are described qualitatively in regulatory review documents (National Marine Fisheries Service 2001a; Tetra Tech 2004; North Pacific Fishery Management Council (2004). While these industry costs represent real losses to society, they are not closely related to the so-called revenue at risk. Methods do exist for estimating the costs of fishery time and area closures based on extensions of the Random Utility Model (RUM; McFadden 1981). RUM was initially developed to model transportation mode choice (Ben-Akiva and Lerman 1985; Domencich and McFadden 1975). Early applications to natural resources focused on estimating demand for recreational fisheries and associated non-market values (Bockstael 1989). RUM was first extended to commercial fisheries by Bockstael (1983), and has increasingly been used to model spatial economic decisions in fisheries (Dupont 1993; Holland and Sutinen 2000). Haynie and Layton (2004) estimated a spatial choice model for groundfish trawl fishing in the Bering Sea, establishing an initial milestone toward quantifying the cost of critical habitat closures. In this project we addressed two specific limitations of the standard approach used by Haynie and Layton (2004) – generality and usefulness to management. First, Haynie and Layton (2004) did not address costs for a large offshore fishing fleet. The second, and more fundamental, limitation arises from the imposition of an unrealistic choice set on fishers. Haynie and Layton (2004) divided the fishing ground into 29 geographic areas, based on statistical reporting, of which 18 contained most of the fishing. This artificial division is unlikely to correspond to actual fleet choices. The Bering Sea/Aleutian Islands area is a huge expanse with a complex coastal and subsurface geography; a realistic choice set for the trawl fleet would contain a much larger set of more precise locations. The theoretical justification for RUM requires that the choices represent a complete set of discrete, independent, and available alternatives (McFadden 1981). Empirical applications of RUM are appropriate to the extent that they model a realistic decision-making problem for individual agents. Like the  Economic Valuation Of Critical Habitat Closures, Berman et al.  4  applications to recreational fishing, commercial fishing applications work best when they involve specific alternative fisheries and discrete alternative fishing sites (Berman, Haley, and Kim 1997). We therefore aimed to develop a realistic method of valuing habitat-driven fishery closures. To be useful to managers, the method must satisfy four criteria. It must: 1.  Be consistent with RUM, benefiting from RUM's theoretical and practical advantages;  2. Differentiate among a large number of small areas distributed over a large geographic space, so that it is relevant to decisions regarding marine mammal critical habitat; 3. Recognize costs of reduced fishing flexibility to an at-sea processing fleet as well as the shorebased fleet; 4. Provide estimates of impacts on fisheries linked directly to ecological variables that are consistent with habitat models for SSLs and other protected species potentially interacting with fisheries. In this way, estimated fisheries values can be compared directly to habitat requirements, both of which may vary over time. This work could therefore be considered a test of two primary hypotheses about Bering Sea and Gulf of Alaska fisheries: 1.  Data on measured and modeled environmental variables can predict spatial variation in the density of catchable fish biomass at the small temporal and spatial scales relevant to realistic modeling of fishing fleet choice sets and management needs; and  2. Resulting predictions of fish biomass, along with data on prices and indicators of fishing costs, can predict spatial choices of the shore-based and offshore fishing fleet in a way that can be used to derive profit functions under the assumptions of RUM. Of course, closing marine habitat to fishing may benefit fisheries in the long run, enabling higher future catches outside the boundaries of the closed areas as stocks rebuild. These benefits could be significant, but estimating such benefits is outside the modest scope of this project. We also did not attempt to estimate the value of protected species saved through closures to fisheries. We addressed only the shortterm cost to the fishery of foregone fishing opportunities – the cost that often poses the main obstacle to the creation of such reserves. Our goal was to design and demonstrate a method to quantify the net cost to the fishing industry of closing areas to fishing that satisfies the above four criteria. Our intent was to improve existing economic models of spatial choice in fisheries by relaxing unrealistic restrictions on spatial decision-making while incorporating detailed and flexible geographic scales. The research plan included four specific objectives: •  Develop and test a scientifically defensible method to value commercially fished areas at flexible temporal and spatial scales relevant to management decisions;  •  Demonstrate a specific application of the method by estimating the cost to Bering Sea and Gulf of Alaska groundfish trawl fisheries of changes in Steller sea lion critical habitat closures;  •  Create maps of relative fisheries values for comparison to maps of relative importance to SSL recovery, in order to assist management decision-making under uncertainty;  •  Generalize the method to evaluate fishery time and area closures for any protected species, or for marine conservation generally.  Economic Valuation Of Critical Habitat Closures, Berman et al.  5  METHODS Theoretical approach For this study, cost of habitat protection means the opportunity costs, or profits foregone, from time and area closures and gear restrictions. Valuing this cost started with a model of fishing fleet decision-making consistent with the assumptions of RUM. The RUM has been widely used to model spatial economic decisions in fisheries (Dupont 1993; Holland and Sutinen 2000). Its advantages include the ability to model choices among multiple spatial alternatives, straightforward computation using maximum likelihood techniques, and direct derivation of welfare estimates under a reasonable set of assumptions. We extended the RUM approach to address the goals of the project by making the following five assumptions: 1.  The probability of use of each alternative (when it is open to fishing) is based on the RUM;  2. Modeled alternatives are small geographic units with similar fish habitat; 3. Expected catch in each alternative unit depends on predicted fish density times geographic area; 4. Since alternatives are very small in relation to the total fishery area, the probability that any vessel uses a given area during each fishing day is small (generally < 1%); 5. A large number of vessel-days per month are observed in each modeled fishery (generally > 100). Under these assumptions, the number of landings in an area during a specified time can be approximated by the Poisson probability distribution. Using probabilistic models of count data to approximate the RUM is new for commercial fisheries, but analogous to an approach proposed by Guimares et al. (2003) to model siting decisions regarding industrial facilities. Since the underlying choice probabilities conform to the assumptions of RUM, we may invoke it to estimate the value of each small choice area from the estimated parameters of the profit function, following Small and Rosen (1981). Berman (2006) describes the specific technical approach developed under the project for implementing the extension of RUM to commercial fisheries, consistent with the above five assumptions. Suppose the utility that an agent in group i derives from selecting choice j at occasion k, is  U ijk = Vijk + [ηijk + ε ijk ]  (1)  where Vijk represents the profit for alternative k, ηijk is a random term with a zero mean whose distribution may be correlated with observed data, and ηijk is a random term with a zero mean with an  independently and identically distributed type one extreme value distribution. The random variable η may, for example, model systematic but unobserved differences in operating costs or other information that varies among vessels in a fishery. For a given value of ηijk, the conditional probability πijk that a fisher from group i chooses area j ∈ Jk is given by:  log π ijk = αVijk + η  ijk  -γ 'ik  (2)  Where  γ ik = log  ∑  j∈J k  αVijk  e  + ηijk  (3)  Under the five assumptions listed above, the conditional probability for the number of vessels yijk from group i observed catch in area j during occasion k may be approximated by a Poisson distribution:  Economic Valuation Of Critical Habitat Closures, Berman et al.  6  prob( yijk = y ) = λijk y exp(−λijk ) / y !  (4)  where  λ jk = nk exp( x jk β + ηijk − γ 'k )  (5)  The unconditional probability distribution for yjk depends on the distribution of η. However, equations (4) and (5) represent a form of "overdispersion" in the Poisson model (Cameron and Trivedi 1986). For example, if eη is assumed to have a gamma distribution, the Poisson approximation to the mixed logit becomes a negative binomial model, whose parameters can be estimated easily with conventional maximum likelihood. Since the underlying choice probabilities conform to the assumptions of RUM, we may invoke RUM to estimate the conditional value of an area from the estimated parameters δ, ε, and gk following Small and Rosen (1981). Given that the vessel has chosen to take a fishing trip, the ex-vessel price, and the relevant geographical information, the difference in value between two subsets J1k and J2k of the choice set Jk is related to the parameter γ'ik:  S J1k − S J 2 k = −(nk / α ) log ∑  j∈ J1 k  αVijk  e  + ηijk − (nk / α )∑  j∈ J 2 k  αVijk  e  + ηijk  (6)  The opportunity cost of closing area j to fishing during choice occasion k reduces to:  −nk log[1/(1 − π jk )] / α  (7)  Since the expected value of ηijk = 0, a point estimate of the opportunity cost may be derived by evaluating the coefficients of a negative binomial regression for Equation (2). However, the complexities that Hensher and Greene (2003) outline for derivation of welfare estimates with mixed logit models apply here. In general, bootstrapping is necessary to generate confidence intervals around the point estimate.  Data sources This was a data-intensive project. Although we did not generate new data from field observations, much of the work on the project involved processing primary data, including environmental indicators, fish catch and effort, fisheries openings and habitat regulations, and indicators of prices and costs. Spatial resolution: A key issue that had to be determined at the start of the project was the identification of the appropriate spatial and temporal scales for the analysis. Three criteria drove the decision: realistic behavioural choices of fishing fleets, management needs, and the resolution of the available environmental data. We decided to analyse the data at multiple scales, with different types of analyses at each scale. The three scales selected were: 1.  The Alaska Department of Fish and Game (ADF&G) statistical area (statarea). Each statarea measures one-half degree of latitude by one degree of longitude (roughly 30 x 40 km), with the areas further subdivided in coastal areas around landforms. In any given year, about 700 of approximately 1,700 statareas report some groundfish fishing activity. The statarea was used for testing the feasibility of the approach (Berman 2006), but the available data support more detailed spatial scales, which also better represent fishing decisions and satisfy management needs;  2. A 9 x 9 km2 grid, covering the entire Gulf of Alaska and Bering Sea regions. This geography was selected as the finest scale that was supported when considering the resolution of the available satellite data. There were approximately 45,000 9 x 9 km2 grid cells within the U.S. Exclusive Economic Zone (EEZ) in the Bering Sea and Gulf of Alaska; and  Economic Valuation Of Critical Habitat Closures, Berman et al.  7  3. Within the Gulf of Alaska, we analysed the data on a 3 x 3 km2 grid. This downscaling was possible for the Gulf of Alaska because of the availability of oceanographic model output (described below). The area contained within the model area inside the U.S. EEZ contained about 36,000 3 x 3 km2 cells. Environmental indicators: Environmental indicators used in the project included bathymetry, remote sensing data, and oceanographic model output. Data sources on fish biomass density and fishing effort provided some limited additional environmental measurements, as described below. Remote sensing data were obtained for four indicators: sea surface temperature, sea surface height, wind, and chlorophyll-a. Each indicator has its own strengths and weaknesses. Temperature is a basic habitat indicator for most species. However, cloudiness disrupts satellite measurements, and the limitation of remotely sensed temperature to surface waters compromises the utility of this indicator for bottomdwelling fish. Sea surface height is measured as an anomaly from mean sea level. Sea surface height is affected by salinity, temperature, and persistent atmospheric pressure anomalies, all of which could indicate habitat variation. However satellite measurements of sea surface height tend to produce inaccurate readings near shallow coastlines affected by tidal action. Wind, inferred from satellite readings of wave height, indicates surface mixing and could be an important indicator of surface enrichment. However, interference with land causes wind data to be unavailable in coastal areas where significant fishing takes place. Chlorophyll-a provides a measure of primary production. However because it is derived from colour images, clouds and low light conditions in winter combine to yield few if any valid readings in December and January. Sea ice also affects both colour and radar satellite measurements. We derived monthly climatologies from available remotely sensed data, beginning in 1993, except for chlorophyll data, which began in 1997. For the Gulf of Alaska, Al Hermann and Dave Musgrave provided output from a Regional Ocean Modeling System (ROMS) model developed for the Gulf of Alaska (Hermann et al. 2002; Hermann and Stabeno 1996). Model outputs were provided at a 3 x 3 km2 resolution, summarized as 2-week averages, for all of calendar year 2001. ROMS model indicators included temperature, salinity, and velocity vectors in three directions at 30 different vertical levels. In addition, the model also calculated a mixed layer depth and sea surface height anomaly. Each of the three types of environmental indicators has advantages and disadvantages. The ROMS output has the ability to ‘see’ the ocean in three dimensions, as well as infer ocean dynamics through currents and eddies. However, it is difficult to validate the model to assess its accuracy at the fine scale we used. (For example, modeled and remote-sensed sea surface height are positively correlated, but the correlation coefficient is only 0.3, so we included both indicators for the 3 x 3 km2 Gulf of Alaska analyses.) ROMS model output was available only for a portion of the Gulf of Alaska (GOA). Remotely sensed data provide direct measurements of dynamic environmental conditions. However, as discussed above, sea ice, clouds, and low light conditions reduce the spatial and temporal extent of these data. Bathymetry provided the best quality and coverage, and served as an important but limited habitat indicator. For the statistical analyses, we extracted four levels from the ROMS output: surface, bottom, and the levels immediately above and below the mixed layer depth. We calculated horizontal velocity from the two horizontal vectors. In addition, to reduce collinearity of variables at different depths, we represented the surface and bottom values as differences from the level below the mixed layer. Remotely sensed data generally supported monthly climatologies at 7 – 10 km resolutions. After selecting a minimum usable quality level based on available quality flags, we interpolated the remote sensing data to common 9 x 9 km2 (entire study area) and 3 x 3 km2 (GOA) grids. Metadata (Gregr and Coatta 2008) contain the technical specifications for the sources and processing of the environmental data. After developing the five data sets of environmental climatologies, we generated slopes (rates of change across space) for bathymetry, sea surface temperature (SST), and sea surface height (SSH). The formula (using SSH as an example), was : sshslope = (180/3.14159)*arctan{ [((SSHn-SSHs)/2) 2 + ((SSHw-SSHe)/2)2 ]1/2 } The subscripts n, s, e, w denote the adjacent cells to the north , south, east and west, respectively.  (8)  8  Economic Valuation Of Critical Habitat Closures, Berman et al.  Since bathymetry is measured in metres, and horizontal distances in kilometres, we divided the formula result by 1000 to obtain bottom slope in degrees. Slopes for SST and SSH represent fronts that could indicate areas of unique habitat. We used linear interpolation as needed to map the 12 monthly climatolgies onto 26 2-week intervals for the 3 x 3 km2 GOA study area. Data on fish density and fishing effort: We derived data measuring spatial fish density and fishing effort from three sources. ADF&G landings data provide information on effort and catch (including bycatch and discards) delivered to shore-based processors at the spatial resolution of the ADF&G statarea. Data provided by ADF&G include round weight and number of boats, summarized by species, gear, port landed, and month for two calendar years (1998 and 2002). Complete data are available when data represent more than three vessels in an observation. ADF&G also indicated which statareas in a given month had at least one but less than three vessels delivering a given species, gear, port, month combination. We combined all trawl and all fixed gears to increase the number of data points. Since completing a fish ticket is voluntary for the offshore sector, we used only onshore landings in the statistical analysis. ADF&G data were used for testing the methods for mapping spatial fisheries values (Berman 2006), but had insufficient spatial resolution to value fisheries closures related to SSL critical habitat. Alaska Fisheries Science Center personnel kindly provided NMFS trawl biomass survey data for the Gulf of Alaska in 2001 (Alaska Fisheries Science Center 2001). NMFS uses the survey for area-wide stock assessments. The survey records represent individual trawl hauls at over 500 points, providing considerable spatial variation. However, the survey had limited spatial and temporal coverage. All data points lie west of Middleton Island in the Central Gulf. The survey was taken in late spring and early summer, starting in mid-May and ending in mid-July. The survey moved systematically from west to east. Trawl depths for the survey were limited to less than 600 metres. Despite these limitations, the NMFS survey data do provide a spatially random sample taken with standard gear over a large geographic area. The survey data included 421 points that fell within the ROMS model spatial coverage. In addition to haul weight by species and haul duration, each haul also recorded time and location, surface temperature and gear temperature. The fishery catch and effort data used for the 9 x 9 km2 study came from the NMFS fisheries observer program. NMFS made available to the project individual haul records for the North Pacific groundfish fishery from 1993 through 2003. Geographic coverage for the observer data extends to the entire Alaskan fishing grounds. The data include all gears and all species caught, including bycatch and discards. For 2001, approximately 44,000 hauls were observed, including 100 percent of hauls in the Bering Sea and about 30 percent of the hauls in the Gulf of Alaska. Data on species round weights in each haul were estimated from sampling portions of each observed haul. Most observed hauls included bottom depth and fishing depth as well as time and location. In order to join the different data sets spatially, we related the survey locations and GOA observer trawl ending locations and dates to the ROMS model 3 x 3 km2 grid and two-week time step. We also aggregated all observer data to the 9 x 9 km2 grid and at a monthly time step. Observed catch locations may represent a spatially biased sample of fish density, since the fishing fleet is preferentially targeting (or may be avoiding in the case of unwanted bycatch) areas of the ocean where concentrations of fish are more likely to be found. However, one can test and correct for this bias arising from nonrandom spatial selection at the scale of grid, although not at finer scales. Data on fisheries openings and habitat regulations: Fishing seasons, and in-season time and area closures by gear and sector (inshore vs. offshore where different) were derived from public sources (National Marine Fisheries Service 2001b; 2001c). The primary spatial reference for most in-season fisheries regulations is the NMFS three-digit management area. We also obtained information on seasonal time and area closures and gear restrictions related to bycatch, Steller sea lion critical habitat, and other environmental regulations from applicable sections of the Code of Federal Regulations archived on the NMFS Alaska region website (National Marine Fisheries Service 2001d). NMFS management support personnel provided digital spatial data delimiting the geographic boundaries applicable to each separate environmental regulation. In all, we mapped 78 separate closures, each applying to different gears, fisheries, and time periods. We interpolated where necessary to address spatial and temporal overlap between closed areas and periods and the modeled spatial grid cells and time periods.  Economic Valuation Of Critical Habitat Closures, Berman et al.  9  Indicators of prices and costs: Primary economic factors relating to the value of catch opportunities consisted of targeted fish species prices and the distance traveled to access fishing areas. Ex-vessel prices for trawl and fixed gear landings in the GOA and Bering Sea were obtained from Alaska Fisheries Science Center (2003). For all fisheries, we calculated distance to port as the one-way distance from the catch grid cell to the grid cell of the nearest port used by any vessel of that gear type (trawl or fixed gear), based on ADF&G landings data. Although fixed-gear boats land catch at numerous ports, trawl vessels landed nearly all their catch at eight Alaska ports in 2001: Dutch Harbor, Akutan, King Cove, Sand Point, Kodiak, Kenai, and Cordova. For offshore fisheries, we also considered the distance between consecutive offshore hauls, but rejected that measure in favour of a different approach, which we describe below.  Statistical methods We conducted our analyses at three different spatial scales, using increasingly detailed information about the oceanic environment and fishing-related costs and revenues as we moved from coarser to finer spatial scales. We first estimated monthly results for the Bering Sea and Gulf of Alaska at the resolution of ADF&G statistical areas with data for calendar year 1998. Fishing was relatively unregulated in terms of spatial closures in 1998, with the exception of regulations set by NMFS management areas. After this year, habitat closures, which could not be modeled adequately at this coarse a spatial resolution, increasingly began to affect fishing location choices. We then estimated results for the Bering Sea and Gulf of Alaska at 9 km resolution, using a monthly time step for calendar year 2001. This was the year NMFS adjusted the Steller sea lion regulatory regime to comply with a court order. We had originally hoped to examine these data for additional years, in order to determine the stability of the relationships over the years. However, the complexity of fisheries and environmental regulations, which changed spatially and temporally each year, made this impractical. Finally, we estimated results for the Gulf of Alaska, 2001, at a 3 km resolution using the ROMS output to predict environmental conditions at fishing locations. We estimated equations to explain variation in CPUE as a function of the environmental variables available at each particular scale. For both the 9 and 3 km analyses, data from the NMFS observer program included all species caught in each haul, regardless of whether that species was targeted or not. Consequently, a catch of zero could be inferred for hauls that did not report any catch of a given species. In order to use the zero-landings information to improve predictions of the spatial distribution of different species, we estimated CPUE equations from the observer data using censored regressions (tobit). On the other hand, data available from ADF&G were not available by individual vessel or haul, and included information only from vessels that reported landings of that species. Since it was not possible to infer zero catches from such data, we estimated CPUE for the statareas using ordinary least squares. For all CPUE equations, we obtained a large improvement in the statistical fit by using a loglinear specification, or more precisely (to accommodate the zero observations), the natural logarithm of CPUE + 1. For the 9 and 3 km analyses, we tested and corrected for sample selection bias as necessary to take into account potential correlation between the spatial distribution of variables determining fish distribution and the distribution of fishing effort. To quantify how predicted CPUE, economic variables, and regulatory factors influenced the distribution of fishing effort, we estimated Poisson and negative binomial regressions (Berman 2006). Berman (2006) also illustrated how closely the Poisson regression approximates the multinomial logit model, using example fisheries estimated from the ADF&G statareas. For the 9 and 3 km studies, all equations were estimated using standard maximum likelihood techniques.  10  Economic Valuation Of Critical Habitat Closures, Berman et al.  RESULTS Statistical analysis proceeded sequentially in three stages. First we estimated censored normal (tobit) regression equations explaining the spatial distribution of CPUE in the data source at each time step, using the coefficients to project variation in expected CPUE over the entire study area. Second, we estimated Poisson and negative binomial regressions explaining the spatial distribution of fishing effort at each time step as a function of projected CPUE and economic factors, to obtain a set of spatial profit functions. Third, we used the coefficients from the spatial profit functions to estimate the opportunity cost to the fishery of closing specific areas related to designation of Critical Habitat for Steller sea lions. Berman (2006) discussed the analysis at the level of the ADF&G statarea. This resolution was ideal for developing the theoretical method, and demonstrating the ability of the Poisson regression to approximate closely the multinomial logit equation for the large number of alternatives needed to model site choice realistically. Although Berman (2006) estimated a distribution of value for onshore groundfish fisheries across more than 1,000 statareas in the North Pacific, the spatial resolution of the statarea was too coarse to apply to evaluating the complex spatial geography of Steller sea lion habitat closures, so we do not discuss it further here. Instead, we focus on results obtained for the Bering Sea/Aleutian Islands and Gulf of Alaska (North Pacific) at a 9 km resolution, based on remote sensed environmental data, and for the Gulf of Alaska (GOA) at a 3 km resolution, using output of the GOA ROMS model in addition to the remote sensing data.  Spatial fish density Results from the GOA 2001 bottom trawl survey. Because the trawl survey is a spatially random sample taken during a relatively brief (approximately 2-month) time period, estimating CPUE is straightforward. The spatial sampling is stratified to improve accuracy (National Marine Fisheries Service 2001c). Since our interest is in understanding spatial variation fish density rather than total areawide biomass, we ignored the strata weights in our analysis. The 2001 GOA survey contains 521 sample points, of which 415 observations lie within the boundary of the ROMS model. Nearly all the 106 excluded sample points were in the west end of the GOA (western portion of NMFS regulatory area 610). Figure 1 displays the geographic location of 2001 trawl survey points in comparison to the observed hauls in that year. The 2001 Gulf of Alaska trawl survey was conducted from May 20 through July 23. Oceanographic model outputs were obtained for 14-day periods. The Julian days representing the start of each model period, determined by the correspondence between the Figure 1. Location of 2001 GOA trawl survey available survey data and the ROMS model output, points (orange) and observed haul locations are 137, 151, 165, 179, and 193. (green). Table 1 contains the precise definition of variables included in the 3 km Gulf of Alaska CPUE analysis using the survey data. The dependent variable is the natural logarithm of CPUE, defined as total round weight for each species divided by trawl duration (kg/hour). Individual species considered included pollock, Pacific cod, black cod, and halibut. In addition, we aggregated all flatfish and rockfish species into two species group categories. These represent the main target and bycatch fisheries for groundfish vessels in the GOA. We estimated separate equations for all hauls, and for hauls with average fish weight greater than a given threshold (representing hauls with likely commercial value), derived from the distribution of average fish weights in hauls for each species or species group. Wind data were available for only about two-thirds of the observations. Consequently, we estimated separate equations with and without wind. This yielded four equations for each species: all and large fish, with and without wind.  Economic Valuation Of Critical Habitat Closures, Berman et al.  11  Appendix A contains the complete set of CPUE equation results estimated from the 2001 trawl survey for the Gulf of Alaska study. Equations were estimated as censored normal regressions (tobit), by maximum likelihood. Due to high collinearity of the set of modeled and observed environmental variables, we dropped variables with a probability > 0.3 (absolute value of t-statistic approximately equal to 1) from the equations to increase robustness of the predictions. With separate intercepts for each time period, coefficients should be interpreted as effects of spatial anomalies. In other words, coefficients, except for those for the time periods and constant term, would be identical if all variables were transformed to represent deviations from the respective time-period mean. Since only one observation had wind data in the first time period, the wind equations combine intercepts for the first two periods. The tables in Appendix A display the respective CPUE equations in pairs. The bottom equation on the page is the censored regression; the top equation reports the corresponding ordinary least squares regression for reference. Censoring the equation at a minimum of zero is necessary to avoid generating predictions of negative CPUE, especially in the equations for large fish. Unfortunately, no generally recognized measure of goodness of fit such as R2 exists for the censored equations, which combine a probit and a linear regression. Figure 2 summarizes the results in Appendix A by displaying the level of statistical significance and direction of effect of the set of environmental variables in the multivariate analysis for all fish. Quadratic associations imply an optimal habitat that lies within the range of observations for that variable. The equations fit best for black cod and rockfish, and least well for Pacific cod. The linear regression equations excluding small fish have lower R2 than the corresponding equations for all fish, but not necessarily a worse fit if censoring is taken into account. In general, the equations show a different pattern of significant variables across species, suggesting habitat selection. Several modeled environmental variables such as salinity, temperature at depth and velocity variables have significant estimated effects for each species, suggesting that the ROMS model successfully enables downscaling the CPUE analysis to the 3km scale.  Economic Valuation Of Critical Habitat Closures, Berman et al.  12  Table 1. Environmental variables used for 3 km Gulf of Alaska CPUE equations.  Variable  Definition  Source  constant  constant term (intercept for period 12)  calculated  herm13  dummy variable for two-week period starting with julian day 151  trawl survey date  herm14  dummy variable for two-week period starting with julian day 165  trawl survey date  herm15  dummy variable for two-week period starting with julian day 179  trawl survey date  herm16  dummy variable for two-week period starting with julian day 193  trawl survey date  ldepth  natural logarithm of bottom depth, in metres  trawl survey  l2dep  square of nat. log of bottom depth  calculated  timeldep  julian day times ldepth  trawl survey  lslope  natural logarithm of slope at 3km resolution  NOAA bathymetry  l2slope  square of nat. log of slope  calculated  lmld  natural logarithm of mixed layer depth, in metres  Hermann model  lmld_dep  nat. log. of mixed layer depth times nat. log. of bottom depth  calculated  surftemp  surface temperature  trawl survey  geartemp  gear temperature at fishing depth  trawl survey  gtemp2  square of gear temperature  calculated  tempdep  temperature at first model level below mixed layer depth  Hermann model  mstemp  temperature at depth minus surface temp  Hermann model  bmtemp  bottom temperature minus temperature at depth  Hermann model  lstem  natural logarithm of surface temperature  trawl survey  lgtem  natural logarithm of gear temperature at fishing depth  trawl survey  l2gtem  square of natural logarithm of gear temperature  calculated  lmtem  natural logarithm of temperature below mixed layer depth  Hermann model  lmstem  nat. log of temperature at depth minus nat. log of surface temp  Hermann model  lbmtem  nat. log of bottom temperature minus nat. log of temp at depth  Hermann model  lmsal  natural logarithm of salinity below mixed layer depth  Hermann model  lmssal  nat. log of salinity at depth minus nat. log of surface salinity  Hermann model  lbmsal  nat. log of bottom salinity minus nat. log of salinity at depth  Hermann model  vervelbm  vertical velocity at level below mixed layer depth (10-3m/s)  Hermann model  msvervel  vertical velocity at depth minus vertical vel. at surface (cm/s)  Hermann model  bmvervel  vertical velocity on bottom minus vertical vel. at depth (cm/s)  Hermann model  horvelbm  horizontal velocity at level below mixed layer depth (cm/s)  Hermann model  mshorvel  horizontal velocity at depth minus hor. vel. at surface (cm/s)  Hermann model  bmhorvel  horizontal velocity on bottom minus hor. vel. at depth (cm/s)  Hermann model  ssh  modeled sea surface height, in metres  Hermann model  sshrs  average monthly sea surface height, in metres x 10-2  Remote-sensed  lchla  natural logarithm of chlorophyll, current period  Remote-sensed  lchla1  natural logarithm of chlorophyll, one-period (14-day) lag  Remote-sensed  lchla2  natural logarithm of chlorophyll, two-period (28-day) lag  Remote-sensed  lwind  natural logarithm of average wind speed  Remote-sensed  Economic Valuation Of Critical Habitat Closures, Berman et al.  13  Equations including wind Variable Modeled environment  P. cod  pollock  black cod  halibut  flatfish  rockfish  Bottom depth Bottom slope Mixed layer depth Bottom temp. Temp. gradient Salinity at mld Salinity gradient Vertical velocity Vert. vel. gradient Horiz. velocity Hor. vel. gradient Sea surface height  22  ++  22 +  22 ++  22  22 + 22 -  ++ 22 ++ 2 --  -+ --  Sea surface temp. Sea surface height Chlorophyll-a Lagged chl-a Wind  --  ++ ++ + + -  + + ++  ++ +  ---  +  -  +  +  Remote sensed environment  ++ + +  R sq. OLS  0.28  ++ ++ + -  ++ --  0.33  0.64  0.35  0.30  0.45  P. cod  pollock  black cod  halibut  flatfish  rockfish  Bottom depth Bottom slope Mixed layer depth Bottom temp. Temp. gradient Salinity at mld Salinity gradient Vertical velocity Vert. vel. gradient Horiz. velocity Hor. vel. gradient Sea surface height  22 2  ++  + +  -+ --  22 ++  2  Sea surface temp. Sea surface height Chlorophyll-a Lagged chl-a  --  Equations excluding wind Variable Modeled environment  22 + 22 -  ++ 22 ++ 2 --  22 ++ 2  ++ ---  + + ++  -2 ++ -  ++ +  ---  +  -  +  Remote sensed environment  R sq. OLS  ++ +  ++  0.27  0.33  Significant positive association Significant negative association Significant quadratic association  + 2  ++ --  --  ++  0.58  0.35  Positive association < .01 Positive association < .01 Quadratic association < .01  0.30  0.48 ++ -22  Figure 2. Variables explaining the spatial distribution of fishing effort, Gulf of Alaska, Summer 2001: Equations including wind (top panel) and excluding wind (bottom panel). Red denotes a significant positive association, blue a significant negative association, and purple a significant quadratic effect.  14  Economic Valuation Of Critical Habitat Closures, Berman et al.  Results from the North Pacific 2001 NMFS observer bottom trawl hauls. NMFS observer data provided a large sample of CPUE taken over the entire area and time period used by the fishery. However, unlike the trawl survey, the observer data – indeed any data derived from commercial fishing activity – do not constitute a spatially random sample. One would assume that the fishing fleet preferentially samples areas of the ocean that have target and desired bycatch species present. The fleet may also avoid areas with concentrations of unwanted bycatch species. Not taking this correlation into account could lead to the classic problem of sample selection bias, where the explanatory variables in the relationship predicting CPUE are correlated with the error term. We included all hauls in each CPUE equations, not just those hauls targeting the given modeled species, and ignored the differential rate of observer sampling in the GOA vs. BSAI regions. Including all hauls, rather than just those hauls targeting the given modeled species, increases the sample size and geographic dispersion, as well as reducing the effect of selection bias. However, one must assume that selection may still exist. The simplest method to correct for selection bias is to apply the Heckman (1979) procedure. The Heckman correction involves a two-step process. First one estimates a probit equation for the probability that a grid cell has observed fishing activity. This is essentially a reduced-form equation that combines the variables predicting CPUE with those predicting distribution of fishing effort. From the probit equation, one calculates the inverse Mills ratio (IMR), defined as the ratio of the standard normal density to the cumulative probability for each observation, and includes the IMR as an additional explanatory variable in the CPUE equation. Heckman (1979) discusses how to adjust the standard errors for the second step equation. A more statistically-efficient method would be to estimate the combined probit selection equation and CPUE equation using full information maximum likelihood (FIML). This approach would be preferred in theory where the CPUE equation is a censored regression (observed hauls with zero catch of a given species). In the present case, however, the situation is more complex. The FIML procedure considers the joint probability of the fishing location and presence of a species at that location jointly as a bivariate normal distribution. Applying FIML to estimate CPUE for each species would yield a different set of coefficients for the probit selection equation for each species, based on correlation with the particular CPUE equation. However, the probit equation models the same location choice for all species, so its coefficients should be the same across all species. This more accurately models the process when the species addressed in the CPUE equation represents the dominant target species for the entire fishery (such as pollock in pelagic trawl hauls). On the other hand, applying the Heckman (1979) procedure to a censored regression would estimate the probability that a species is present, along with the CPUE, given the probability of a haul occurring at that location. This more closely models the process when there are many target species represented in the data set of hauls (as is the case for bottom trawl hauls). In theory one could estimate FIML results for the system of equations for all the species combined, along with a single probit selection equation. However, since each species would have its own correlation coefficient with the probit, the FIML multivariate normal equation would be infeasible to estimate in practice.  Economic Valuation Of Critical Habitat Closures, Berman et al.  15  Table 2. Environmental variables for the 9 km Bering Sea/Aleutian Islands and Gulf of Alaska CPUE equations.  Variable  Definition  Source  constant  constant term (intercept for January)  calculated  feb  dummy variable for observation in February  observer haul date  mar  dummy variable for observation in March  observer haul date  apr  dummy variable for observation in April  observer haul date  may  dummy variable for observation in May  observer haul date  jun  dummy variable for observation in June  observer haul date  jul  dummy variable for observation in July  observer haul date  aug  dummy variable for observation in August  observer haul date  sep  dummy variable for observation in September  observer haul date  oct  dummy variable for observation in October  observer haul date  nov  dummy variable for observation in November  observer haul date  dec  dummy variable for observation in December  observer haul date  goa  dummy variable for observation in Gulf of Alaska  observer data  ldepth  natural logarithm of bottom depth, in metres  observer data  l2dep  square of nat. log of bottom depth  calculated  timeldep  month times ldepth  calculated  slope  slope at 3km resolution, in degrees  NOAA bathymetry  2slope  square slope  calculated  sst  surface temperature, degrees C  remote sensed  sst2  square of absolute value of surface temperature  calculated  sstslope  square of sea surface temperature, in degrees  calculated  ssh  average monthly sea surface height, in metres x 10-2  remote sensed  sshslope  slope of sea surface height, in degrees  calculated  lwind  natural logarithm of average wind speed  remote sensed  mwind  natural log. of average wind speed, monthly means for missing values  calculated  mchla  natural logarithm of chlorophyll, monthly means for missing values  remote sensed  mchla1  previous month’s value for mchla  remote sensed  gtimelde  goa times timeldep  calculated  gsst  goa times sst  calculated  gssh  goa times ssh  calculated  glwind  goa times lwind  calculated  gmwind  goa times mwind  calculated  gmchla  goa times mchla  calculated  gmchla1  goa times mchla1  calculated  bottomt  observer haul location (=1 if haul observed; 0 if no haul observed)  observer data  imrx  inverse Mills ratio from applicable probit equation number x  calculated  portdist  distance to port  calculated  xxxtrawl  openings for trawl fishery xxx: pol=pollock, cod=P. cod, atk=Atka mackerel  NMFS regulations  xxxtssl  habitat closures for trawl fishery xxx: same as above, plus mix=other trawl  NMFS regulations  16  Economic Valuation Of Critical Habitat Closures, Berman et al.  Using the 2001 observer haul data, we estimated censored CPUE regressions both with the Heckman method and using FIML for the main target and desired bycatch species for trawl fisheries. As defined above, these are pollock, Pacific cod, Atka mackerel, black cod, and rockfish and flatfish species groups. Where multiple hauls were observed in the same grid cell during the same month, we averaged the CPUE for the respective hauls. We estimated separate equations for winter and summer seasons, where summer is defined as the months of May through October, in order to allow for different behaviour of groundfish during spawning and non-spawning seasons. Table 2 summarizes the exact definition of the explanatory variables available for this analysis, including those needed to estimate the selection probit equations. As in the previous analysis using survey data, the dependent variable is the natural logarithm of CPUE plus one, where CPUE is defined as total weight (extrapolated to the haul from observed samples) divided by haul duration. CPUE units in the observer equations are tonnes per hour. Although the survey CPUE is measured with standard gear, heterogeneity of the trawl fishing fleet could affect measured CPUE. We created a measure of ‘standard CPUE’, defined as the CPUE for a standard, or average-sized boat. We computed standard CPUE in two steps. First, we created an inferred capacity from the maximum haul weight for all species combined, observed for each vessel over the entire year. Then we created standard CPUE by multiplying observed CPUE by the ratio of the average inferred capacity for the fleet (113.8 tonnes) to the inferred capacity of the vessel. For all species, equations for standard CPUE showed a significantly better fit. As discussed in the next section, however, while the environmental data produced a much better fit for the standard CPUE equations for Pacific cod than for the equations for raw CPUE, predictions from the raw CPUE equations provided a much better prediction of the distribution of the Pacific cod fishing fleet. We therefore report both standard and raw CPUE equations for Pacific cod. We included only data from bottom trawl hauls, in order to maintain as much consistency as possible. Since the pollock fishery is primarily a pelagic trawl fishery (exclusively so by regulation in the Bering Sea), we also estimated equations for pollock from pelagic trawl hauls. Pollock CPUE equations estimated for the two gear groups produced very similar results. We therefore report only the bottom trawl results. Appendix B shows the complete estimation results for three sets of equations for each species each season: (1) ordinary least squares using the Heckman (1979) procedure (with corrected standard errors); (2) censored regressions using the Heckman approach, and (3) FIML results. The tables leave out the FIML probit coefficients, which differ slightly for each species. Instead, we show the very similar probit equations for the first stage Heckman selection process. Including all hauls, rather than just those hauls targeting the given modeled species, reduced the effect of selection bias. Indeed, the selection effect, as measured by the coefficient on the IMR, is not significantly different from zero in many of the CPUE equations, and the three specifications often yielded conflicting results regarding the significance and direction of selection bias . Although the selection effect is not robust among the specifications, the three equations generally yielded similar effects for the set of environmental variables. Figure 3 summarizes the results for the censored regressions (second specification) in Appendix B.  Economic Valuation Of Critical Habitat Closures, Berman et al.  17  Winter months Variable Measured environment  pollock  Bottom depth Depth over time Bottom slope  P. cod  A. mackerel black cod  22  22  + 22  + 22  -22  22  22  -  -  +  ++  ++  --  22  rockfish  flatfish  22  22  22  ++ 22  22  22  Remote sensed environment Sea surface temp. SST slope Sea surface height SSH slope  ++  Wind speed Chlorophyll-a  -0.20  Selection bias  ++  --  ++  ++ --  +  Lagged chl-a R sq. OLS  -  -++ -  0.38  0.48  0.32  +  +  --  0.31  0.30 -  Summer months Variable Measured environment  pollock  P. cod  A. mackerel black cod  rockfish  flatfish  Bottom depth Depth over time  22 +  22  22  +  22  22 +  Bottom slope  22  --  22  --  22  22  +  22  22  22  +  ++  SSH slope  +  Wind speed  -  Remote sensed environment Sea surface temp. SST slope  +  Sea surface height  ++  Chlorophyll-a  --  -  ++  Lagged chl-a  -  --  -  --  0.24  0.23  0.39  0.36  0.50  0.29  -  --  --  ++  -  R sq. OLS Selection bias  Significant positive association  +  Positive association < .01  Significant negative association  2  Positive association < .01 Quadratic association < .01  Significant quadratic association  ++ -22  Figure 3. Variables explaining spatial distribution of fishing effort, Bering Sea/Aleutian Islands and Gulf of Alaska, 2001: Winter months (top panel) and Summer months (bottom panel). Red denotes a significant positive association, blue a significant negative association, and purple a significant quadratic effect. Source: Appendix B  18  Economic Valuation Of Critical Habitat Closures, Berman et al.  The observer CPUE equations fit best for Atka mackerel in winter and rockfish in summer, and least well for pollock in both seasons, and Pacific cod in summer. One might expect that the habitat indicators would have less ability to explain location and density of more mobile species. The equations differ significantly between summer and winter for most species, especially with respect to the direction and significance of the remote-sensed variables. Since the environmental variables change strongly between summer and winter, as noted in means and standard deviations displayed with the equation results, the different equation coefficients do not necessarily imply that the fish have moved far between summer and winter habitats.  Spatial distribution of effort The aim of the statistical analysis is to derive a profit function – that is, a relationship for Vijk (Equation 2) based on associating the spatial distribution of observing fishing effort with a set of factors representing spatial variation in fishing costs and revenues. We hypothesized that spatial variation in predicted CPUE – whether derived from survey data or from the fishing fleet as a whole – would be highly correlated with spatial variation in revenues from fishing. Important spatial cost factors (discussed above) include distance to port and fishery regulations creating time and area closures and gear restrictions. Each statistical relationship that predicts the spatial distribution of CPUE for a given species can potentially generate an estimated equation for the spatial distribution of effort. We first discuss results derived from the Gulf of Alaska 2001 bottom trawl survey, and then examine results estimated using CPUE predictions derived from fisheries observer data. Results from the GOA 2001 bottom trawl survey. In practice, limits on spatial and temporal scales and geographic extents in available predictions of CPUE determine the scale and boundaries of the analyses that can be performed for spatial distribution of effort. The Gulf of Alaska 2001 bottom trawl survey generates profit functions at a very fine spatial scale (3 x 3 km2 grid). However, our analyses using the trawl survey are limited in spatial extent by the geographic coverage of the GOA oceanographic model since the CPUE predictions rely on the model output. For the GOA, we also limited the choice set to the bottom depths included in the survey (less than 600 m), which also approximates the deepest trawl haul observed in the GOA fisheries. The time frame for the survey data collection also limited the GOA effort analysis. The 2001 survey was conducted during the summer months – late May through late July. The only fisheries open for a substantial portion of the survey period were the sablefish and halibut longline fisheries, and some flatfish trawl fisheries. Rockfish trawling was open during a three-week period in July. To estimate equations relevant to the pollock and Pacific cod trawl fisheries – the largest GOA fisheries and the ones mainly affected by Steller sea lion critical habitat designations – we had to project the CPUE equations past the end of the range of the data used to estimate those relationships. We hypothesized that the relationship of the environmental variables to CPUE estimated during the survey period held for the duration of trawl fishing in 2001 (late August to late October). Examining whether an equation estimated from early summer data predicts the distribution of fishing effort in fall, controlling for other relevant factors, tests the hypothesis of stability of the relationship, providing a mechanism to validate the overall method. One could in theory also project the CPUE equations forward into the early spring and winter months. However, we doubt that such an out-of-season prediction would be valid, due to possible behavioural changes in groundfish during the spawning season. Whether or not the equations are used to predict CPUE beyond the time horizon of the data used to estimate the relationships, generating predictions for the entire Gulf of Alaska involves projecting outside the range of the independent variables observed in the roughly 400 survey sample locations. This is basically a problem of boundary conditions. The range of modeled temperature and salinity, as well as remote-sensed variables such as chlorophyll-a and sea surface temperature, extends beyond the range of these variables in sampled areas. This is particularly a problem in estuaries and other coastal environments. The problem is magnified by the fact that the CPUE equations are estimated as loglinear equations (which yields a much better fit), so the errors, after converting to CPUE units, are exponential. One way of handling the data range problem is to limit the CPUE predictions to the range of the observations on the independent variables within the sampled points. This approach would be appropriate for handling out-of-sample prediction within the time horizon of the survey data set. Unfortunately, such  Economic Valuation Of Critical Habitat Closures, Berman et al.  19  an approach becomes problematic when extending predictions over time, as the ranges of the variables change seasonally. We opted instead to use the predictions generated by the equations for large fish, and censor a small percentage of high CPUE predictions. In essence, this approach assumes that the CPUE equations generate good predictions of suitable habitat for mature fish, but cannot predict high abundance locations within that habitat. The rationale is based on the scale of spatial aggregations of fish compared to the spatial scale of the data. Within any 9x9 km2 area representing a grid cell, we expected that the trawl survey would find the species present in significant numbers if environmental conditions favour it, but would only randomly find large aggregations within suitable habitat. The fishery, on the other hand, will search for such aggregations within the local area. Given the censoring approach, there remains the question of how to choose the limit for each fishery. We determined the censor with a stepwise process based on the log likelihood of the equation for the distribution of fishing effort. We reduced the upper limit of predicted CPUE by one integer level of predicted natural logarithm of CPUE at a time, until the log likelihood stopped increasing. This occurred for pollock at a value of 7, and at a value of 3 for Pacific cod. One test of the validity of such an approach is whether the profit functions it generates appear reasonable. Table 3 shows the best-fitting negative binomial regressions for the distribution of GOA shore-based pollock and Pacific cod trawl fishing effort during fall 2001, as a function of censored CPUE predictions generated from the 2001 bottom trawl survey. The shore-based pollock and Pacific cod fisheries are the main GOA fisheries affected by the final Steller sea lion regulations, which went into effect on July 17 of that year. Because average CPUE changes each time period, often along with fisheries and habitat regulations, a separate intercept term is required for each time period to represent the overall value of fishing opportunities during the period: the parameter γ'k in Equation (5). The constant term represented the intercept for model period 19 (August 23-September 5), the first summer period during which any landings were recorded for either fishery. Coefficients for the other periods represented effects relative to period 19. Pacific cod trawling occurred in the GOA in period 19, but was closed for the duration of period 21. Fishing ended for both fisheries on October 31 (period 23). The negative binomial equations in Table 3 exhibit a high degree of dispersion: the variance scale factor is around 100: that is, the variance is 100 times the mean. The high variance creates convergence problems for the algorithm, as small changes in the scale factor have little effect on the log likelihood or the other equation coefficients. The variance multiplier for pollock had to be approximated by estimating the equation with a fixed value for the scale factor, changing the value until the log likelihood stopped increasing. Nevertheless, the equation results for both pollock and Pacific cod appeared reasonable. The coefficients on expected censored CPUE are positive and significant, and the coefficients on distance to port are significant and negative. We excluded grid cells from the choice set that were subject to regulatory closures during the entire model period. For pollock, some hauls occurred in areas that were open for a portion of the period. The coefficients on the regulatory variables – the fraction of the time included in a fishery opening and the fraction of the time subject to a habitat closure, respectively – have the expected signs and are highly significant. For Pacific cod, hardly any hauls were observed in such partially closed areas, so no coefficients could be estimated. GOA directed trawl fisheries besides pollock and Pacific cod include rockfish and flatfish fisheries. These fisheries are covered under a diverse set of regulations and are exempt from most, but not all the Steller sea lion habitat regulations. A distinguishing characteristic of these fisheries is the dependence on retained bycatch for valuable species – primarily black cod, but also rockfish for some vessels when the directed fishery is closed – for additional revenue. For the shore-based fishery in particular, it is often difficult to determine what the target species is for a given vessel based on an individual observed haul. Consequently, we estimate a single equation for ‘other’ mixed trawl fisheries.  20  Economic Valuation Of Critical Habitat Closures, Berman et al.  Table 4 shows the best-fitting negative binomial regressions for the distribution of fishing effort estimated for GOA other trawl fisheries. For the shore-based fisheries, large rockfish, black cod, and flatfish all significantly predict location choice. Using the same censoring procedure as before, the best fit is achieved at a censor of 3 for rockfish, 4 for flatfish and black cod, and 6 for pollock. The constant term represents period 12; the intercepts for the remaining periods represent effects relative to period 12. Port distance is again negative as expected, and highly significant. Pollock is abundant in the GOA, but of lower value, especially for smaller fish. The coefficient on all pollock is negative and significant. With the variety of desired and unwanted species, a tradeoff between expected revenue and CPUE for other onshore trawl fisheries similar to those displayed in Figure 4 was not straightforward to construct. However, the coefficient on port distance was similar to those for the other two shore-based fisheries. Several offshore trawl fisheries operate in the Gulf of 5 Alaska. All GOA pollock is allocated to the shore-based sector, and too 4 little Pacific cod is allocated to the offshore sector to estimate a choice 3 equation. No mother ships operated in the GOA in 2001, but a number of trawl catcher-processors prosecuted 2 the rockfish and flatfish fisheries. During the period in summer that 1 rockfish was permitted as a target species (model periods 15 and 16), 0 enough hauls (279) were observed 0 6 12 18 24 30 36 42 48 54 60 66 72 78 84 90 to estimate a choice equation for Distance to port (km) offshore fishing effort. The primary Summer/fall Pacific cod Summer/fall Pollock catch target for most of these hauls appeared to be rockfish. However, Figure 4. Profitability tradeoffs implied by the distribution of black cod, which has a high market fishing effort: Gulf of Alaska 2001. value, remains an important bycatch species. Distance to port is not important for GOA offshore fisheries, since no portion of the Gulf lies particularly far from at least one port with ability to accommodate a mid-sized trawl catcher-processor. As mentioned above, we considered the distance between consecutive offshore hauls as a measure of travel cost. While highly significant when included in the equation, such an approach to measuring travel cost is of little practical value, for two reasons. First, modeling distance between hauls requires constructing a separate choice set for each grid cell in which a haul is observed, since that cell becomes the starting point from which to measure distance to the next haul. One would have to construct thousands of choice sets, which, even with sampling, would involve millions of observations in order to model the fishery. In other words, it negates the advantage of the count models for partial aggregation of the large spatial choice set. Second, knowing that a cell was selected because it was close to a previous haul location provides insufficient information for predicting spatial values. One would need to model the set of consecutive hauls in a nested choice framework. In particular, one might assume a nested choice model, where the probability of selecting a grid cell is a function of that cell’s CPUE and the “inclusive value” (McFadden 1981), representing the value of the set of future opportunities from the series of consecutive future hauls. The inclusive value at each level depends on the value of the unknown parameter for the inclusive value at a lower choice level (corresponding to the next haul). With dozens of consecutive hauls during a fishery, and with thousands of choices for each haul, the likelihood function would be infeasible to compute, even if one could construct the set of inclusive values data set. Given these challenges, we decided to experiment with a simple approach that addressed travel distance conceptually in a nested model, but in a way that was easy to understand conceptually and straightforward to estimate. The inclusive value at each stage (haul) is a function of CPUE, the distance to a subsequent haul, and the subsequent haul’s inclusive value. If travel distance between hauls is typically low, much of the variance in inclusive value will be captured with a measure of average CPUE in the neighbourhood around the grid cell in question.  Economic Valuation Of Critical Habitat Closures, Berman et al.  21  Table 3. Negative binomial regressions for distribution of Fall 2001 Gulf of Alaska onshore pollock and Pacific cod trawl fishing effort: Maximum likelihood estimates (t-statistics in parentheses). The dependent variable was the number of target fishery hauls observed. pollock Constant Period 20  Pacific cod  -8.791 ***  -4.517 ***  (-6.35)  (-11.0)  1.929 ***  --  (2.64) Period 21 Period 22  0.9192 (0.79)  (-5.17)  -1.110 ***  0.6378 ***  (-4.84) Period 23  -0.5618 ** (-2.05)  Fishery regulatory opening  -2.281 ***  (2.61) -0.6534 ** (-2.02)  5.5433 *** (3.91)  Fishery SSL habitat closure  -1.4517 *** (-5.39)  Expected survey CPUE  0.0007414 *** (3.02)  Distance to port (km)  -0.01665 *** (-14.36)  Variance Scale  100 --  0.07873 *** (3.49) -0.01223 *** (-7.46) 136.28 *** (6.42)  Log-likelihood  -898.5455  -875.1  Initial Log-Likelihood  -1350.877  -1312.6  904.7  875.0  105,205  93,219  Likelihood ratio squared Observations ** p < 0.05 *** p < 0.01  Economic Valuation Of Critical Habitat Closures, Berman et al.  22  Table 4. Negative binomial regressions for distribution of Summer and Fall 2001 Gulf of Alaska onshore and offshore ‘Other’ trawl fishing effort: Maximum likelihood estimates (t-statistics in parentheses). The dependent variable was the number of target fishery hauls observed.  onshore Constant Period 13 Period 14 Period 15 Period 16 Period 17 Period 19 Period 21 Period 22 Period 23 Expected CPUE, rockfish Expected CPUE, flatfish Expected CPUE, black cod Expected CPUE, pollock Expected neighbourhood CPUE, rockfish Distance to port (km) Variance Scale Log-likelihood Initial Log-Likelihood Likelihood ratio squared Observations *p < 0.1 ** p < 0.05 ***p < 0.01  offshore  -5.6947 (-4.92) -1.7916 (-1.86) -1.9221 (-1.86) 0.25172 (0.28) 0.37629 (0.56) 0.05481 (0.08) -0.04711 (-0.07) -3.1653 (-1.35) -0.08453 (-0.12) -1.6386 (-1.53) 0.09464 (1.73) 0.03951 (2.15) 0.02056 (1.69) -0.01313 (-2.07)  ***  -6.6377 *** (-25.67)  * *  0.3170 (1.59)  *  0.02694 *** (3.06)  ** *  2.082E-04 * (1.83)  ** 0.06038 ** (2.94)  -0.01282 *** (-5.40) 50 --308.0 -2840 5064 *** 22,403  150 --946.5 -1246 599.7 *** 42,435  Economic Valuation Of Critical Habitat Closures, Berman et al.  23  In the hauls observed from the North Pacific trawl fisheries in 2001, travel distance between hauls was less than 30 km about 80% of the time. We constructed a measure of rockfish neighbourhood CPUE that was a simple average of the predicted CPUE of all grid cells within 30 km of each cell. Cells within this range that were on land or otherwise unavailable for fishing – for example, with water depths deeper than 600 m – were included in the average with a zero value. The second column of coefficients in Table 4 shows the results of estimating the negative binomial equation for spatial choice for the GOA offshore ‘Other’ trawl fishery. The constant term for the offshore fishery represents period 15; the intercept for period 16 represents an effect relative to period 15. The censoring procedure produced the highest log likelihood at a value of 4 for the natural log of rockfish CPUE, 8 for black cod, and 3 for the 30 km rockfish neighbourhood CPUE. The coefficient on rockfish CPUE, black cod bycatch, and the rockfish neighbourhood CPUE are all positive and significant. Results for the 2001 North Pacific trawl fisheries at 9 km resolution. Scaling up to the North Pacific and moving from the oceanographic model output to remote-sensed observations as the basis for predicting CPUE required several additional factors to be addressed. In particular, the offshore fisheries in the Bering Sea are large and complex. The Atka mackerel fishery is an important offshore fishery that is specifically included in the Steller sea lion habitat regulations. A number of large mother ship fleets operate in the Bering Sea. However, observers on mother ships cannot determine exactly where the haul was caught, and many hauls could have occurred in a grid cell other than the one recorded for the mother ship when the haul was brought aboard. Consequently, we consider only catcher-processors to represent the offshore fleet. The larger catcher processors that operate in the Bering Sea can trawl to deeper levels. We therefore used 700 m as the maximum depth. We did not censor the CPUE predictions, since they derive from equations estimated from the fisheries with the same data set. The larger grid cell size, however, limits the flexibility of the neighbourhood CPUE calculation. Neighbourhood CPUE for the offshore fisheries in the 9 km analysis averaged CPUE in all grid cells within 20 km of the target cell, and excluded the central cell from the calculation. In addition, we used data for all 12 months of the year, so the choice equations included hauls for the entire period in 2001 during which directed fishing for that target species was permitted. Tables 5 and 6 show the negative binomial regression results for the distribution of fishing effort for 2001 Bering Sea and Gulf of Alaska shore-based and offshore trawl fisheries, respectively. We considered pollock, Pacific cod, Atka mackerel (offshore only), and rockfish target fisheries. The fishing regulations for the diverse Bering Sea flatfish fisheries are so complex that we did not attempt to estimate equations for this fishery, but rather used rockfish as a basis from which to generalize to the other trawl fisheries. The constant term represents the month of January; the other months represent effects relative to January for each month where hauls were observed. In addition to separate intercepts for each time period, we used a separate intercept for GOA grid cells to represent the effects of a difference in the observer coverage, as well as differences in fishery regulations in the Gulf of Alaska. Where a t-statistic is missing for the variance scale factor, the scale was estimated iteratively, as discussed above. Partial months’ fishery openings and habitat closures (as discussed above for the GOA) again significantly explain the distribution of effort for pollock, as well as the offshore Atka mackerel fishery. Offshore pollock fishing is prohibited within the Catcher Vessel Operating Area (CVOA). The positive, significant coefficient for the CVOA suggests a congestion effect, or some pathway by which competition with the offshore fishery affects profitability of the onshore fleet.  Economic Valuation Of Critical Habitat Closures, Berman et al.  24  Table 5. Negative binomial regressions for the distribution of 2001 Bering Sea and Gulf of Alaska shorebased trawl fisheries maximum likelihood estimates (t-statistics in parentheses). Dependent variable: number of target fishery hauls observed.  pollock Constant  -1.2498 * (-1.90)  P. cod -4.6277 *** (-6.14)  rockfish -6.8094 *** (-4.25)  February  2.4079 ** (2.39)  2.0584 ** (2.19)  March  3.6583 *** (3.33)  5.4063 *** (4.73)  4.3474 ** (2.34)  3.4613 *** (3.56)  4.5357 ** (2.35)  April  -0.74609 (-0.69)  May  1.231 (0.97)  2.112 * (1.75)  June  -2.2732 *** (-2.69)  July  -0.4371 (-0.42)  0.26946 (0.31)  2.9205 (1.30)  August  -2.6464 * (-1.80)  -0.6889 (-0.42)  2.8411 (1.33)  September  -2.5728 * (-1.66)  0.62073 (0.62)  3.7546 ** (2.03)  October  -3.0826 * (-1.95)  0.29461 (0.18  2.3453 * (1.90)  GOA  -2.5295 *** (-4.48)  -0.9924 (-0.83)  GOA*February GOA*March  -0.95673 (-1.09) -1.2965 (-1.12)  0.59191 (0.46)  2.587 ** (2.24)  -0.6407 (-0.40) -4.5917 ** (-2.10)  -3.2224 ** (-2.07)  GOA*April  -1.6775 (-1.04)  GOA*July  -0.0331 (-0.02)  Economic Valuation Of Critical Habitat Closures, Berman et al.  25  Table 5 (continued). Negative binomial regressions for the distribution of 2001 Bering Sea and Gulf of Alaska shore-based trawl fisheries.  pollock  P. cod  GOA*August  4.1665 ** (2.51)  GOA*September  3.5723 ** (2.12)  0.11671 (0.08)  GOA*October  3.3818 ** (2.13)  1.9185 (0.98)  Fishery opening  0.46144 (0.32)  Habitat closure  -7.4125 ** (-2.32)  Catcher Vessel Op. Area  rockfish -2.4645 (-1.25) -1.9815 (-1.29)  3.2991 ** (2.59)  Expected target CPUE  0.94094 *** (3.62)  1.0686 *** (2.75)  0.54686 *** (2.72)  Distance to port (km)  -0.01415 *** (-6.40)  -0.00984 *** (-6.36)  -0.01252 *** (-7.42)  Variance Scale  22.052 *** (10.17)  108.43 *** (5.23)  38.552 *** (4.51)  Log-likelihood Initial Log-Likelihood Likelihood ratio squared Observations *p < 0.1 ** p < 0.05 ***p < 0.01  -1029.8 -18566 35073 *** 14,504  -363.89 -6594.6 12461 *** 15,800  -375.82) -3831.8 6912.0 *** 18,732  Economic Valuation Of Critical Habitat Closures, Berman et al.  26  Table 6. Negative binomial regressions for the distribution of 2001 Bering Sea and Gulf of Alaska offshore trawl fisheries maximum likelihood estimates (t-statistics in parentheses). Dependent variable: number of target fishery hauls observed.  pollock  P. cod CPUE -5.8285 *** (-6.14)  P. cod standard  Constant  -5.3002 *** (-3.42)  -5.4243 *** (-6.08)  February  -3.3406 (-1.40)  1.4846 (1.57)  1.6753 * (1.78)  March  -2.2833 (-0.96)  3.3841 *** (3.16)  3.5669 *** (3.46)  April  -2.6535 (-1.11)  1.4848 (1.20)  1.5528 (1.34)  Atka mackerel  rockfish  -4.8533 *** -2.2217 *** (-3.06) (-8.87) -3.8038 (-1.48)  0.86159 *** (2.95)  2.0219 (1.00)  1.9069 *** (4.68) 2.9953 ** (8.45)  May  -9.195 *** (-3.30)  -0.5648 (-0.41)  -0.5947 (-0.44)  0.54855 (1.29)  June  -4.564 * (-1.90)  0.70685 (0.77)  0.8148 (0.95)  1.0766 *** (3.23)  July  -3.5481 (-1.48)  0.58327 (0.59)  0.61193 (0.68)  1.9931 *** (5.25)  August  -3.4948 (-1.45)  0.87803 (0.85)  0.75062 (0.78)  2.7179 *** (7.19)  -2.463 (-1.03)  0.02562 (0.02)  -0.2418 (-0.21  -3.4555 (-1.44)  0.93725 (0.62)  0.70636 (0.50  -3.7882 ** (-2.55)  -4.0564 *** (-2.92  September October GOA GOA*February  -2.1931 (-1.01)  2.2756 *** (6.76) 2.5314 *** (6.29) -4.2784 *** (-10.28) 1.4902 *** (2.69)  GOA*March  0.41813 (0.57)  GOA*April  2.8132 * 1.95  3.3909 ** 2.59  GOA*May  3.9998 ** 2.40  4.3063 *** 2.65  0.88828 ** (1.98) 4.0211 *** (7.63)  Economic Valuation Of Critical Habitat Closures, Berman et al.  27  Table 6 (continued). Negative binomial regressions for the distribution of 2001 Bering Sea and Gulf of Alaska offshore trawl fisheries.  pollock  P. cod CPUE  P. cod standard  Atka mackerel  GOA*July  1.126 ** (2.02)  GOA*September Fishery opening  -0.9723 (-0.87) 6.9813 * (1.79)  Habitat closure  -6.2714 (-1.29)  Exp. target CPUE  0.91187 *** (4.91)  Neighbour CPUE  0.88797 *** (3.39)  Dist. to port (km)  -0.00521 *** (-12.60)  Variance Scale  Log-likelihood Initial LogLikelihood Likelihood ratio squared Observations  rockfish  18.678 *** (15.02) -2412.55  9.2642 * (1.885)  2.0902 ** (2.45)  0.68025 * (1.75)  0.32935 ** (2.006)  -0.00370 *** (-3.95)  -0.00337 *** (-3.80)  -0.00306 * (-1.853)  100 --  100 --  35 --  -299.26 -3231.2  -306.46 -3231.2  -78.062 -1688.2  1.3225 *** (6.30)  -0.00560 *** (-16.40) 24.25 *** (12.62) -1978.9 -15879  -18440.5 32056 ***  5863.9 *** 16,556  5849.5 *** 16,556  3220.33 *** 370  27799 *** 17,824  11,832 *p < 0.1 ** p < 0.05 ***p < 0.01  The coefficients on expected target CPUE are all positive and significant. For onshore Pacific cod, we show the equation estimated with raw CPUE as well as the equations for standard CPUE used for the other fisheries. For reasons that we cannot explain, the equation for raw CPUE fits much better for shore-based Pacific cod. The coefficients on distance to port are all negative and highly significant in the onshore fisheries. Interestingly, the coefficients on port distance are also significant and negative for all the offshore fisheries as well. However, the magnitude of the coefficients for the shore-based fishery is about three times as large as for the offshore fishery. Figures 5 and 6 illustrate the tradeoff between distance to port and expected revenues per standard haul implied by the spatial distribution of fishing effort for the onshore and offshore fisheries, respectively. For the onshore fisheries, rockfish and Pacific cod appear to be more sensitive to distance from port than pollock. However, the expected revenue measure assumed average annual ex-vessel prices for pollock. During the winter season, roe adds significant value to the pollock fishery, but we lack reliable data on the implied ex-vessel value of roe. If roe values were included, the slope for pollock would be steeper, implying less, if any, difference between pollock and the other fisheries. The slopes of the profitability tradeoffs implied by the choice equations are much lower across the board than those for the onshore fisheries. The scales of the two figures are identical to facilitate the comparison. The offshore fisheries include an additional target species, Atka mackerel. The Atka mackerel target fishery is a geographically and temporally compressed fishery that is heavily constrained by Steller sea lion  28  Economic Valuation Of Critical Habitat Closures, Berman et al.  regulations. The combination of fisheries and habitat regulations left relatively few cell-months available for fishing, so results for this fishery should be considered less reliable than those for the other fisheries. Nevertheless, the tradeoffs shown in Figure 6 suggest that the estimated profit functions for Atka mackerel imply similar economic tradeoffs between remoteness and expected CPUE as the other offshore fisheries. The neighbourhood CPUE variable is positive and highly significant for the offshore pollock fishery. Indeed, the coefficient on the neighbourhood expected CPUE does not differ significantly from the coefficient on the cell’s expected CPUE. The neighbourhood CPUE did not explain the distribution of effort for the other three offshore fisheries modelled. Since nearly half of all consecutive hauls took place within 9 km of the previous haul, the results suggest that the spatial scale might be too coarse to pick up the neighbourhood effect for these less mobile fisheries. Instead, the coefficient for expected CPUE for the target cell includes whatever neighbourhood effect may exist.  Figure 5. Profitability tradeoffs implied by the distribution of fishing effort: 2001 Onshore trawl fisheries.  Spatial values and the cost of habitat protection The negative binomial regression equations (Tables 3 through 6) generated estimates of the spatial distribution of fishery values for the various target fisheries and fleet sectors. The equations estimated for the GOA at a 3 km resolution may be considered more accurate representations of the spatial relationships, but are more limited to the GOA summer fisheries. The 3 km GOA results best illustrate the finescales effect of regulations and effects of changes to boundaries of individual closure units. The 9 km results best illustrate the overall impact of the Endangered Species Act Critical Habitat program as a whole on the North Pacific groundfish fleet.  Figure 6. Profitability tradeoffs implied by the distribution of fishing effort: 2001 Offshore trawl fisheries.  To assess the impact of the Steller sea lion habitat closures, one must first establish a baseline for comparison. Since the North Pacific groundfish fisheries are regulated under a total allowable catch (TAC) regime, one may make the assumption that any expansion or contraction of fishing opportunities would result in only a slight change in the number of hauls made by the fleet during the year. Instead, there would be a redistribution of a fixed level of fishing effort and catch among available fishing locations. Consequently, one may assume that habitat regulations do not change the number of hauls, and focus on estimated changes in the value per haul.  Economic Valuation Of Critical Habitat Closures, Berman et al.  29  The equations for fishing location choice were estimated based on the fisheries openings and habitat closures in effect during 2001, so the 2001 regime would ordinarily become the baseline for the analysis. However, the Steller sea lion regulatory regime actually changed twice during 2001, first in June 10, due to the expiration of the congressionally mandated delay in implementing the judicial order requiring a revision to the Reasonable and Prudent Alternatives (RPA), and again on July 17 when NMFS completed the regulations to comply with the order. This created an unstable baseline that complicated the analysis. To address this moving baseline, we estimated the spatial distribution of relative values by evaluating Equations (6) and (7) assuming that fishery TAC and bycatch regulations remained as they were in 2001, but that all Steller sea lion habitat regulations were removed. Hypothetically removing the Critical Habitat fishing restrictions inflates the estimated total value of the fishery above that actually realized by the fleet under the habitat regulations in effect during 2001. However, it provides a constant baseline with which to compare changes in relative value caused by the imposition of the habitat regulations. GOA 3 km resolution. As mentioned above, trawl fishing was effectively closed in the GOA in the summer of 2001 until July 17. Figure 7 shows the estimated baseline net value of summer fishing in the Gulf of Alaska in 2001, assuming that all Steller sea lion Critical Habitat Closures were removed. We computed the estimates in the table by evaluating Equation (7) with the coefficients shown in Tables 3 and 4, and applying a scale factor unique to each fishery. To scale the values, we first converted the scale from expected trawl survey CPUE to expected catch Figure 7. Net value of summer 2001 Gulf of Alaska trawl fisheries. per fishery haul by multiplying by ratio of average catch of the target species in trawl fishery hauls to predicted survey CPUE for the grid cells where hauls were observed. This adjusted for the difference between survey CPUE and expected fishery catch. We then multiplied by average exvessel prices to convert the units from tons of fish per haul to dollars per haul. Finally, we scaled up to total fishery value by multiplying the result for each month by the number of hauls observed during that month in summer 2001, divided by the estimated percentage of hauls in the GOA sampled by observers.1 The estimates in Figure 7 suggest that the mixed trawl offshore fishery is more profitable than any of the onshore fisheries. However, the mixed trawl fishery includes values for sablefish bycatch, for which reliable ex-vessel price data do not exist. Using the estimates in Figure 7 as a base, it is possible to estimate the opportunity cost to the summer and fall 2001 fisheries of the Steller sea lion habitat closures that came into effect on July 17. Figure 8 shows the estimated costs for the three shore-based trawl fisheries, as well as for the offshore rockfish trawl fishery. The numbers are computed by evaluating Equation (6) with J1k representing the  Figure 8. Estimated cost per haul of Steller sea lion closures for onshore Gulf of Alaska trawl fisheries, summer and fall, 2001.  According to the National Marine Fisheries Service, “The portion of the catch sampled by observers varies by region, vessel-type, gear-type, and target fishery. Since 2001, vessels with observers in the BSAI have accounted for approximately 90% of the groundfish tonnage caught and observers have sampled the catch from about three-fourths of the hauls/sets. Vessels with observers in the GOA have accounted for approximately 40% of the groundfish tonnage caught and observers have sampled the catch from about twothirds of the hauls/sets.” (http://www.afsc.noaa.gov/FMA/spatial_data.htm)  1  30  Economic Valuation Of Critical Habitat Closures, Berman et al.  choice set that would have been available if fishing had been permitted within the Steller sea lion closure areas, and J2k equals the choice set actually available for each fishery and time period. Since we only have an approximate figure for the estimated number of hauls for each target fishery, Figure 8 shows the cost in terms of dollars per haul. The estimated cost for Pacific cod shore-based trawl fishery is quite large: around $1,800 per haul. The estimated cost for the pollock fishery is about half, but still substantial. These are the two main fisheries affected by the Steller sea lion closures. The estimated costs for the other onshore and offshore fisheries are small. The Steller sea lion regulations closed relatively few areas of the Gulf of Alaska to all trawl fishing, and those areas – such as three nautical mile no fishing zones around rookeries and haulouts – were of relatively little importance to these fisheries. Figure 9 shows the relative importance of the Steller sea lion closure areas compared to the areas remaining open by expressing the costs in terms of the percentage of baseline values. The figures suggest that the closures cost the Pacific cod trawl fishery about 38% of the profit per haul, and the pollock fishery about 28%. The cost to the other trawl fisheries was slight: 0.2% of the operating profits. The numbers in Figure 9 are measured as a percentage of the potential profit, defined as the baseline profits in Figure 7. The baseline, as discussed above, represents a somewhat arbitrary standard. Nevertheless, the results suggest that the cost to the two fisheries specifically targeted for regulation to protect Steller sea lion foraging around rookeries and haulouts in the GOA is substantial, while the cost to other trawl fisheries is slight. The spatially detailed analysis for the GOA provided an opportunity to measure the change in the cost to the fisheries resulting from relatively small changes in the boundaries to the closed areas. We chose the Chiniak Gully research area to illustrate the capabilities of the method. The Chiniak Gully closure was a relatively small area of the GOA off Kodiak Island that was closed to all trawling during the month of August 2001 so that NMFS could conduct research on the effect of fishing on localized depletion of fish stocks. Figure 10 shows the estimated cost of the August 2001 research trawl closure, using the methods described. The cost is measured in terms of the average cost per haul during the portion of the closure period that the respective target fishery was open to trawl fishing in the GOA. Again, the estimated cost is highest for the Pacific cod fishery: nearly $200 per haul. It turns out the Chiniak Gully was a highly profitable area for the Pacific cod trawl fishery during this time, as it had both high expected CPUE values and was close to the major fishing port of Kodiak. The cost to the pollock fishery was much less, but still significant. The cost for the other shore-based trawl fisheries was relatively small. In this case, the other most valuable trawl target fisheries in the Chiniak Gully area (rockfish) was closed in the GOA. The trawl  Figure 9. Estimated cost of Steller sea lion closures as percentage of profit for onshore Gulf of Alaska trawl fisheries, summer and fall, 2001.  Figure 10. Estimated cost per haul of August 2001 Chiniak Gulley trawl closure for onshore Gulf of Alaska trawl fisheries.  Figure 11. Estimated cost of August 2001 Chiniak Gulley trawl closure as percentage of profit for onshore Gulf of Alaska trawl fisheries.  Economic Valuation Of Critical Habitat Closures, Berman et al.  31  sablefish bycatch allowance had also already been reached. While flatfish trawling was still permitted, estimated profitability for this target fishery was much lower in the GOA in summer and fall. To illustrate this point, Figure 11 shows the costs of the Chiniak Gully closure expressed as the percentage change in the value per haul. The percentage costs are much more closely aligned. According to these estimates, the research trawl closure cost the trawl fisheries about 2-4 percent of profits, with the lowest percentage cost falling on the pollock fishery, and the highest on the Pacific cod fishery. North Pacific, 9 km resolution. We projected CPUE for the 9 km study of the North Pacific as a whole from equations estimated from the trawl fishery itself, unlike the 3 km GOA results which were derived independently from the NMFS trawl survey. However, CPUE for each species was estimated from the groundfish fishing fleet as a whole, using data on all bottom trawl hauls from all fisheries rather than just from hauls targeting that particular species. This means that the projections of CPUE from the statistical equations do not represent the magnitude of target fishery hauls likely to be realized by any particular component of the fishery prosecuting a target species, even though we believe the relative CPUE estimates are valid. For example, the expected CPUE from a standard bottom trawl haul is likely to underestimate expected catches of very large pollock catcher-processors, and over-predict CPUE of smaller Pacific cod catcher vessels. Because of the selection bias discussed above, there is no simple procedure for adjusting the expected standard bottom trawl haul CPUE for all fisheries to the expected target fishery haul. This created a scale problem that made it difficult to estimate a baseline value, as presented for the GOA in Figure 7, or even a cost per haul, as described for Figure 8.  Net value (% change)  With Atka mackerel, we have the additional problem of a small number of hauls and limited geographic dispersion, making the estimated equation imprecise. However, since the method of estimating changes in net fisheries values relies on evaluating the difference in a logarithmic function (Equation 6), changing the scale factor for the base value has only a slight effect on the resulting change in value. Consequently, we showed the estimated costs of Steller sea lion habitat closures in terms of the percentage change in value per haul. Scaling issues with assessing the percentage change in values are basically limited to uncertainty in weighting of GOA values relative to the generally much larger BSAI values.  Figure 12. Estimated percentage change in net value of North Pacific trawl fisheries with three Steller sea lion habitat protection regimes.  As mentioned, the Steller sea lion regulatory regime changed twice in 2001: first when the Congressional moratorium expired on June 10, and again on July 17, when NMFS completed work on the new regime. We assessed the effect of the changing regulatory environment on the fisheries by estimating the relative fisheries values associated with a set of successive regulatory regimes. The hypothetical baseline regime  Economic Valuation Of Critical Habitat Closures, Berman et al.  32  involved no Steller sea lion habitat closures, but retained trawl fishery closures associated with other management objectives (for example, crab bycatch), as well as the general fishery TAC and bycatch-related closures. We defined the first regime as the set of restrictions that related to all trawl fisheries. Most were in place before NMFS designated Critical Habitat for Steller sea lions. We defined Regime 2 as the set of regulations in place on January 1, 2001, and Regime 3 as the set of regulations that went into effect on July 17, 2001. Only minor adjustments to the Steller sea lion regulations have occurred since 2001. Trawl fishing was effectively closed between June 10 and July 17, so we made no attempt to value that intervening period. Table 7 shows the results of evaluating Equation (6) for the three regimes. The results appear quite different from those shown for the GOA. Figure 12 illustrates the relative magnitude of the changes for different fisheries and regulatory regimes. Table 7. Estimated cost of three regimes of Steller sea lion protective measures as a percentage of net profit per haul: Bering Sea/Aleutian Islands and Gulf of Alaska trawl fisheries, 2001.  Fleet  Regime 1 (%)  Regime 1 to regime 2 (%)  Regime 2 to regime 3 (%)  Cumulative, three regimes (%)  Pollock catcher boats  1.0  18.1  -3.9  15.2  Pollock catcher-processors  0.2  2.0  0.3  2.5  P. cod catcher boats  0.6  4.8  5.9  11.3  P. cod catcher-processors  0.4  4.6  2.8  7.8  Atka mackerel catcher-processors  0.5  3.0  3.8  7.3  Other trawl catcher boats  1.1  --  --  1.1  Other trawl catcher processors  0.04  --  --  0.04  Our estimates suggested that the largest cost of the Steller sea lion regulations for the North Pacific fisheries has fallen on pollock catcher boats, as a result of the shift from Regime 1 to Regime 2. The revised regulations in summer 2001 actually mitigated some of that cost for the onshore pollock fleet, while increasing the cost substantially for the Pacific cod and Atka mackerel fleets. The costs to the offshore fleets are smaller across the board than the costs estimated for the shore-based fleets. In particular, the costs for the offshore pollock fleet appear relatively small: almost an order of magnitude smaller than the costs to the shore-based fleet. One important regulatory closure areas affecting the offshore pollock fleet is the large Steller Sea Lion Conservation Area (SSLCA) in the southeastern Bering Sea. In 2001, SSLCA catch limits did not affect pollock catcher-processors. Pollock fishing within the SSLCA closed early for the mother ship sector in the pollock A season, which might have led to a substantial estimated cost. However, as discussed above, we cannot estimate a reliable equation for mother ships, so we cannot estimate the cost to that sector of the fleet.  Economic Valuation Of Critical Habitat Closures, Berman et al.  33  DISCUSSION This report presented an exploratory study designed to develop a new set of methods for estimating spatial values of ocean fisheries at fine spatial scales, and to apply the methods to estimate the cost to the North Pacific groundfish fisheries of spatial closures. As an exploratory study, the criteria for success should be somewhat different from what they might be for a more applied study that was designed to provide findings more readily applicable to management decisions. The two general hypotheses – that environmental conditions explain and can predict the spatial distribution of fish density, and that predicted spatial fish densities predict the distribution of fishing effort at fine spatial scales – received strong empirical support. The estimates of the cost of habitat closures developed from the statistical analyses appear plausible. The cost estimates varied widely for different sectors of the trawl fleet, ranging, for all North Pacific fisheries, from a 10-15% loss of profits for pollock and Pacific cod catcher boats, to negligible or modest for pollock catcher processors and the rockfish and flatfish fisheries. Estimated costs for Atka mackerel and Pacific cod catcher-processors lay midway between the two extremes. Using a different and spatially more precise data set for estimating CPUE in the GOA, we found that the estimated regulatory costs to the Pacific cod and pollock shore-based trawl fisheries were even greater: about twice as high as for the North Pacific. The question that arises is how confident we can be that these estimates reflect true conditions for the fishing fleet. As discussed above, confidence intervals can only be derived from bootstrapping. In this case, bootstrapping is very cumbersome, since we have in essence a three-stage statistical procedure. The CPUE equations for the North Pacific involved a two-stage sample selection bias correction. The predictions then feed into a third-stage probabilistic count model, which itself was only an approximation, albeit one likely to be accurate, for a multinomial logit model that underlies the derived welfare estimates. Since our goal was to demonstrate the method rather than derive numerically precise values, we did not attempt such a computationally complex procedure to generate confidence intervals, and leave this for further research. Nevertheless, the empirical results for the different statistical analyses, combined with general characteristics of the RUM approach, provided some insights into the likely nature and magnitude of the uncertainty in the estimates. First, the coefficient on CPUE in the negative binomial regressions becomes a multiplicative scale factor in Equations (6) and (7) for the estimated values. The total values and values per haul are inversely proportional to the magnitude of this coefficient. Confidence intervals around this coefficient are quite precise (t-statistics around 3) for all fisheries except offshore Pacific cod and Atka mackerel. Estimated costs for these fisheries were in the intermediate range. However, the coefficient on CPUE scales total values estimated both with and without the habitat regulations. The estimated values are relatively insensitive to the magnitude of the coefficient (Figure 7). A much bigger source of uncertainty in the estimates derived from uncertainty in the predictions of CPUE itself. The equations explaining the distribution of pollock and Pacific cod spatial densities were generally less precise than those of the other fisheries. Further work to improve the ability to explain and predict pollock and Pacific cod spatial distribution would have priority for improving the accuracy of the estimates of the cost of fisheries closures. Another question that arises is whether the statistical results and the cost estimates that follow from them represent an artifact of special conditions present in 2001, or whether the relationships are stable over time. It was our original intent to test the stability of CPUE predictions across a number of years. The requirement to compile, analyze, and code the extremely complex and continuously-changing spatial regulatory environment for the North Pacific fisheries in order to correct for selection bias in the CPUE equations made it impossible for us to conduct such stability tests. Clearly, this remains a high priority for future research. As an exploratory study, the refinements discussed here should be undertaken before applying the results in a management context. Once questions about the stability and robustness of the results are resolved, the analysis can be applied directly in a management context to estimate the costs and benefits of all proposed regulatory changes that involve time and area closures for the groundfish fleet. The analysis of the Chiniak Gully research closure provided a simple example of the kind of information that the analysis can provide to management and the public.  34  Economic Valuation Of Critical Habitat Closures, Berman et al.  CONCLUSIONS The project aimed to design and demonstrate a method to quantify the net cost to industry of closing protected areas to fishing that (1) could take advantage of RUM’s theoretical and practical advantages; (2) could be applied at a spatial scale relevant to decisions regarding marine protected areas; (3) included estimates of costs of reduced fishing flexibility to an at-sea processing fleet as well as the shore-based fleet; and (4) provided estimates of fisheries impacts linked directly to environmental variables relevant to habitat models for Steller sea lions and other protected species potentially interacting with fisheries. Our goal was to improve existing economic models of spatial choice in fisheries by relaxing unrealistic restrictions on spatial decision-making while incorporating detailed and flexible geographic scales. We successfully addressed the four stated objectives. We developed and tested a scientifically defensible method to value fishery use areas at flexible temporal and spatial scales relevant to management decisions. The method produced predictions of relative value across the entire U.S. EEZ in the North Pacific at a detailed spatial scale in different seasons. The method could easily be generalized to evaluate fishery time and area closures for any protected species or for marine conservation generally. Finally, we demonstrated an application of the method to estimate cost to Bering Sea and GOA groundfish trawl fisheries of changes to Steller sea lion critical habitat closures. The method generated plausible statistical results that distinguished relatively costly regulations from those involving relatively modest or negligible costs, and the relative burdens on different sectors of the fishing fleet. This approach provides management with credible independent estimates of the effect of regulations on net profits for the first time. While the actual magnitude of the estimated costs may remain uncertain, the percentage changes in costs are likely to be reliable. Further research will be required to demonstrate the robustness and stability of the estimated relationships over time, as well as to compute bootstrapped confidence intervals around estimates of values and costs. The main weakness of the method is its requirement for managing and analyzing large volumes of spatially explicit data. The constraint is both a problem of computational resources and one of human resources. Taking the next step towards implementing the method as a standard practice for NMFS as part of the regulatory review process would require that the agency make an institutional commitment to developing the analytical capability to manage and process large environmental and regulatory datasets. While the agency certainly has this capability, it has in the past been deployed to other research and management tasks.  ACKNOWLEDGEMENTS The authors acknowledge the large number of individuals and organizations who provided data and assistance with its interpretation. Remote-sensed data for the study were obtained from a variety NOAA and NASA public internet data distribution sites. The Alaska Fisheries Science Center provided trawl survey data for the Gulf of Alaska. The NOAA Alaska Region provided data from the fisheries observer program. Dave Musgrave and Al Hermann provided unpublished oceanographic model output developed at NOAA PMEL. Steve Lewis provided geo-spatial data on fishery regulations from the NOAA Alaska Region. The North Pacific Universities Marine Mammal Research Consortium and the North Pacific Marine Science Foundation provided financial support for this and related research. Finally, the authors acknowledge many individual staff members of the North Pacific Fishery Management Council, Alaska Fisheries Science Center, PMEL, and representatives of the fishing industry for useful comments and feedback on the study design and preliminary results.  Economic Valuation Of Critical Habitat Closures, Berman et al.  35  REFERENCES Alaska Fisheries Science Center. 2001. 2001 Gulf of Alaska biennial groundfish assessment survey, resource assessment and conservation engineering division, AFSC Quarterly Research Reports, July-Sept 2001, http://www.afsc.noaa.gov/Quarterly/jas2001/divrpts_race.htm#2001_GOA_biennial. Alaska Fisheries Science Center. 2003 North Pacific groundfish stock assessment and fishery evaluation reports. http://www.afsc.noaa.gov/refm/stocks/assessments.htm. Ben-Akiva, M., and Lerman, S. 1985. Discrete choice analysis: Theory and application to travel demand. Cambridge, MA: MIT Press. Berman, M. 2006. Modeling spatial choice in ocean fisheries. Marine Resource Economics, 21: 387-406. Berman, M., Haley, S., and Kim, H. 1997. Estimating net benefits of reallocation: Discrete choice models of sport and commercial fishing. Marine Resource Economics 12(4): 307-327. Berman, M., and Sumaila, R. 2006. Discounting and amenity values in ecosystem restoration. Marine Resource Economics 21(2): 211-219. Bockstael, N., McConnell, K., and Strand I. 1989. A random utility model for sportfishing: Some preliminary results for Florida. Marine Resource Economics 6: 245-260. Bockstael, N., and Opaluch, J. 1983. Discrete modelling of supply response under uncertainty: The case of the fishery. Journal of Environmental Economics and Management 10: 125-137. Bucke, D., 1989. Histology. In: Austin, B., Austin, D.A. (Eds.), Methods for the Microbiological Examination of Fish and Shellfish. Wiley, New York, pp. 69-97. Cameron, A., and Trivedi, P. 1986. Econometric models based on count data: comparisons and applications of some estimators and tests. Journal of Applied Econometrics, 1: 29-54. Curtis, R., and Hicks, R. 2000. The cost of sea turtle preservation: The case of Hawaii's pelagic longliners. American Journal of Agricultural Economics 82(5): 1191-97. Domencich, T.A., and McFadden. D. 1975. Urban travel demand: A behavioural analysis. Amsterdam: North Holland. Dupont, D.P. 1993. Price uncertainty, expectations formation, and fishers' location choices. Marine Resource Economics 8: 219-247. Gaugh, Jr., H.G., 1992. Statistical Analysis of Regional Yield Trials. Elsevier, Amsterdam. Gregr, E.J., and Coatta, R. 2008. Environmental data for the eastern North Pacific and Bering Sea. Fisheries Centre Research Reports 16(6). 76 pp. Gregr, E.J., and Trites, A. W. 2004. Niche overlap between Alaskan Steller sea lions and trawl fisheries. Poster presented at the 22nd Lowell Wakefield Fisheries Symposium, Anchorage, Sep. 30-Oct. 3. Guimaraes, P., Figueirdo, O., and Woodward, D. 2003. A tractable approach to the firm location decision problem. Review of Economics and Statistics 85(1): 201-204. Haynie, A., and Layton, D. 2004. Estimating the economic impact of the Steller sea lion conservation area: Developing and applying new methods for evaluating spatially complex area closures. IIFET 2004 Japan Proceedings (International Institute of Fisheries Economics and Trade). Corvallis, Oregon. Heckman, J. 1979. Sample selection bias as a specification error. Econometrica, 47: 153-161. Hensher, D.A., and Greene, W.H. 2003. The mixed logit model: The state of practice. Transportation 30(2): 133-76. Hermann, A.J., Haidvogel, D. B., Dobbins, E. L., and Stabeno, P. J. 2002. Coupling global and regional circulation models in the coastal Gulf of Alaska. Progress in Oceanography. 53: 335-367. Hermann, A. J. and Stabeno, P. J. 1996. An eddy resolving model of circulation on the western Gulf of Alaska shelf: Model development and sensitivity analyses. Journal of Geophysical Research 101: 1129-1149. Holland, D., and Sutinen, J. 2000. Location choice in New England trawl fisheries: Old habits die hard. Land Economics 76(1): 133149. McFadden, D. 1981. Econometric models of probabilistic choice. In C. Manski and D. McFadden (Eds.) Structural analysis of discrete data: With econometric applications. Cambridge, Mass. MIT Press, pp 198-272. National Marine Fisheries Service. 2001a. Steller sea lion protection measures final supplemental environmental impact statement, United States Department of Commerce, National Oceanic and Atmospheric Administration, National Marine Fisheries Service, Alaska Region, November 2001. National Marine Fisheries Service. 2001b. Status of 2001 groundfish fisheries - by gear type. http://www.fakr.noaa.gov/2001/status.htm. National Marine Fisheries Service. 2001c. 2001 Gulf of Alaska Biennial Groundfish Assessment Survey. http://www.afsc.noaa.gov/Quarterly/jas2001/divrpts_race.htm.  36  Economic Valuation Of Critical Habitat Closures, Berman et al.  National Marine Fisheries Service. 2001d. 2001 Steller sea lion protection measures and groundfish harvest specifications. http://www.fakr.noaa.gov/sustainablefisheries/2001hrvstspecssl.htm. North Pacific Fishery Management Council. 2004. Regulatory impact review/regulatory flexibility analysis for amendments 84/76/19/11/8 to the BSAI groundfish FMP (#84), GOA groundfish FMP (#76), BSAI crab FMP (#19), scallop FMP (#11), and the salmon FMP (#8) and regulatory amendments to provide habitat areas of particular concern. Initial Draft, September, 2004. Roberts, R.J., 1993. Ulcerative dermal necrosis (UDN) in wild salmonids. In: Bruno, D.W. (Ed.), Pathological conditions of wild salmonids. Fisheries Research: 17: 3-14. Small, K., and Rosen, H. 1981. Applied welfare economics with discrete choice models. Econometrica. 49: 105-130. Stewart, D.A., Agnew, D., Boyd, R., Briggs, R., Toland, P., 1993. The derivation of changes in Nephrops per unit effort values for the Northern Ireland fishing fleet. Fisheries Research 17: 273-292. Tetra Tech FW, Inc. 2004. Draft environmental impact statement for essential fish habitat identification and conservation in Alaska, Appendix C, regulatory impact review/initial regulatory flexibility analysis. National Marine Fisheries Service, January, 2004.  Economic Valuation Of Critical Habitat Closures, Berman et al.  37  APPENDICES APPENDIX A. EQUATIONS FOR SPATIAL DISTRIBUTION OF CATCH PER UNIT OF EFFORT (CPUE), ESTIMATED FROM THE SUMMER 2001 NMFS GULF OF ALASKA BOTTOM TRAWL SURVEY  38  Economic Valuation Of Critical Habitat Closures, Berman et al.  1. Equations with Wind (only shown if absolute value of t for wind > 1) Pacific cod, average weight > 0.5 kg Limited Dependent Variable Model - CENSORED regression Ordinary least squares regression. Dep. Variable LBPCOD Observations 264 Weights ONE Mean of LHS 0.1617331E+01 Std.Dev of LHS 0.2134344E+01 StdDev of resid. 0.1898167E+01 Sum of squares 0.8791417E+03 R-squared 0.2662060E+00 Adj. R-squared 0.2090663E+00 F[ 19, 244] 0.4658862E+01 Prob value 0.3991631E-08 Log-likelihood -0.5333954E+03 Restr.(b=0) Log-l -0.5742529E+03 Amemiya Pr. Criter. 0.3875997E+01 Akaike Info.Crit. 0.4192389E+01 ANOVA Source Variation Deg. freedom Regression 0.3189352E+03 19. Residual 0.8791417E+03 244. Total 0.1198077E+04 263. N(0,1) used for significance levels. Variable Coefficient Std. Error t-ratio Prob. Var. Mean Constant 3651.8 1333. 2.739 0.0062 J162 2.2328 0.8727 2.559 0.0105 0.2765 J177 2.5571 1.467 1.743 0.0814 0.2500 J192 3.6429 2.045 1.781 0.0749 0.2462 LDEPTH 21.023 5.532 3.800 0.0001 4.8944 L2DEP -1.9598 0.5609 -3.494 0.0005 24.2132 LSTEM -3.1445 1.299 -2.420 0.0155 2.3164 LGTEM 33.528 21.26 1.577 0.1148 1.7553 LSLOPE 0.35871 0.3821 0.939 0.3478 3.2087 L2SLOPE -0.69028E-01 0.7476E-01 -0.923 0.3559 11.2445 LMTEM -0.64826 4.017 -0.161 0.8718 2.0037 LMSTEM 19.805 6.670 2.969 0.0030 -0.0902 LMSAL -2138.6 776.6 -2.754 0.0059 3.4659 LBMSAL -46.985 13.30 -3.532 0.0004 0.0383 BMHORVEL 0.51526E-01 0.5817E-01 0.886 0.3757 2.8017 SSHRE 0.36187E-01 0.1089 0.332 0.7397 -3.4680 LWIND 1.4795 1.103 1.342 0.1796 1.7965 L2GTEM -8.2877 5.829 -1.422 0.1550 3.1014 L2MSAL 306.44 113.3 2.705 0.0068 12.0132 LMLD_DEP -0.12687 0.1491 -0.851 0.3949 7.2535 ***************************************************************************** Limited Dependent Variable Model - CENSORED regression Maximum Likelihood Estimates Log-Likelihood.............. -363.32 Threshold values for the model: Lower 0.0000 Upper ********** N(0,1) used for significance levels. Variable Coefficient Std. Error t-ratio Prob. Var. Mean Constant 11000. 3022. 3.640 0.0003 J162 5.4903 2.000 2.745 0.0061 0.2765 J177 7.8656 3.386 2.323 0.0202 0.2500 J192 11.843 4.843 2.445 0.0145 0.2462 LDEPTH 82.904 18.82 4.404 0.0000 4.8944 L2DEP -7.9522 1.898 -4.189 0.0000 24.2132 LSTEM -8.4387 2.728 -3.093 0.0020 2.3164 LGTEM 159.31 61.83 2.577 0.0100 1.7553 LSLOPE 2.7343 1.409 1.941 0.0523 3.2087 L2SLOPE -0.48996 0.2570 -1.906 0.0566 11.2445 LMTEM -14.693 9.276 -1.584 0.1132 2.0037 LMSTEM 44.992 15.32 2.936 0.0033 -0.0902 LMSAL -6532.9 1765. -3.702 0.0002 3.4659 LBMSAL -109.08 30.66 -3.558 0.0004 0.0383 BMHORVEL 0.14673 0.1244 1.179 0.2384 2.8017 SSHRE 0.29768 0.2418 1.231 0.2182 -3.4680 LWIND 2.9177 2.369 1.232 0.2181 1.7965 L2GTEM -39.537 16.70 -2.367 0.0179 3.1014 L2MSAL 942.04 257.4 3.660 0.0003 12.0132 LMLD_DEP -0.38099 0.3333 -1.143 0.2530 7.2535 Sigma 3.3130 0.2491 13.300 0.0000  Mean Square 0.1678607E+02 0.3603040E+01 0.4555426E+01 Var. st. dev. 0.4481 0.4338 0.4316 0.5090 5.0025 0.1872 0.1431 0.9759 5.3610 0.1970 0.0553 0.0278 0.0242 2.9325 1.6767 0.1697 0.5184 0.1912 3.0037  Var. st. dev. 0.4481 0.4338 0.4316 0.5090 5.0025 0.1872 0.1431 0.9759 5.3610 0.1970 0.0553 0.0278 0.0242 2.9325 1.6767 0.1697 0.5184 0.1912 3.0037  Economic Valuation Of Critical Habitat Closures, Berman et al.  39  Pollock, average weight > 0.25 kg Limited Dependent Variable Model - CENSORED regression Ordinary least squares regression. Dep. Variable LBPOLL Observations 263 Weights ONE Mean of LHS 0.8345154E+00 Std.Dev of LHS 0.1541075E+01 StdDev of resid. 0.1293798E+01 Sum of squares 0.4034130E+03 R-squared 0.3516626E+00 Adj. R-squared 0.2951685E+00 F[ 21, 241] 0.6224763E+01 Prob value Log-likelihood -0.4294374E+03 Restr.(b=0) Log-l -0.4864222E+03 Amemiya Pr. Criter. 0.1813936E+01 Akaike Info.Crit. 0.3432984E+01 ANOVA Source Variation Deg. freedom Regression 0.2188139E+03 21. Residual 0.4034130E+03 241. Total 0.6222269E+03 262. N(0,1) used for significance levels. Variable Coefficient Std. Error t-ratio Prob. Var. Mean Constant -4281.5 1321. -3.240 0.0012 J162 0.79908 0.6140 1.301 0.1931 0.2776 J177 0.44705 1.052 0.425 0.6710 0.2510 J192 0.27878E-01 1.474 0.019 0.9849 0.2433 LDEPTH 3.3759 4.608 0.733 0.4638 4.8978 L2DEP -0.49393E-01 0.4374 -0.113 0.9101 24.2447 LSTEM -0.93955 0.8902 -1.055 0.2912 2.3153 LGTEM 33.014 14.54 2.271 0.0232 1.7545 LMLD 4.3831 1.636 2.680 0.0074 1.5058 LMTEM 1.9085 2.834 0.673 0.5007 2.0028 LMSTEM -10.846 5.027 -2.157 0.0310 -0.0900 LMSAL 2462.0 767.5 3.208 0.0013 3.4665 LBMSAL 16.422 9.721 1.689 0.0912 0.0380 HORVELM -0.72220E-01 0.3501E-01 -2.063 0.0391 15.3185 MSHORVEL 0.57074E-01 0.3631E-01 1.572 0.1159 12.8499 BMHORVEL 0.63602E-01 0.4546E-01 1.399 0.1618 2.8098 SSHRE 0.85775E-01 0.7878E-01 1.089 0.2762 -3.4664 LCHLA 0.27958 0.2416 1.157 0.2471 0.7104 LWIND -1.4766 0.7600 -1.943 0.0520 1.7980 L2GTEM -8.5751 4.008 -2.139 0.0324 3.0986 L2MSAL -357.96 111.5 -3.211 0.0013 12.0172 LMLD_DEP -0.60579 0.3453 -1.755 0.0793 7.2726 ***************************************************************************** Limited Dependent Variable Model - CENSORED regression Maximum Likelihood Estimates Log-Likelihood.............. -229.59 Threshold values for the model: Lower 0.0000 Upper ********** N(0,1) used for significance levels. Variable Coefficient Std. Error t-ratio Prob. Var. Mean Constant -11936. 4863. -2.454 0.0141 J162 1.0769 1.963 0.549 0.5832 0.2776 J177 -3.0453 3.449 -0.883 0.3773 0.2510 J192 -5.0420 4.815 -1.047 0.2951 0.2433 LDEPTH 70.138 23.28 3.013 0.0026 4.8978 L2DEP -5.6319 2.157 -2.611 0.0090 24.2447 LSTEM -3.4402 2.771 -1.242 0.2144 2.3153 LGTEM 145.78 78.76 1.851 0.0642 1.7545 LMLD 16.865 7.083 2.381 0.0173 1.5058 LMTEM 18.050 8.915 2.025 0.0429 2.0028 LMSTEM -25.364 16.00 -1.585 0.1129 -0.0900 LMSAL 6797.1 2819. 2.411 0.0159 3.4665 LBMSAL -37.846 34.58 -1.094 0.2738 0.0380 HORVELM -0.16061 0.1148 -1.400 0.1616 15.3185 MSHORVEL 0.12218 0.1142 1.070 0.2848 12.8499 BMHORVEL 0.14484 0.1442 1.004 0.3152 2.8098 SSHRE 0.23584 0.2197 1.073 0.2831 -3.4664 LCHLA 1.5638 0.7075 2.210 0.0271 0.7104 LWIND -3.9803 2.408 -1.653 0.0984 1.7980 L2GTEM -40.774 23.23 -1.755 0.0792 3.0986 L2MSAL -997.94 409.5 -2.437 0.0148 12.0172 LMLD_DEP -2.5134 1.402 -1.793 0.0730 7.2726 Sigma 2.5898 0.2373 10.916 0.0000  0.0000000E+00 Mean Square 0.1041971E+02 0.1673913E+01 0.2374912E+01 Var. st. dev. 0.4487 0.4344 0.4299 0.5069 4.9856 0.1868 0.1428 0.6399 0.1968 0.0553 0.0261 0.0236 11.0040 10.7104 2.9352 1.6797 0.4982 0.1683 0.5174 0.1800 2.9935  Var. st. dev. 0.4487 0.4344 0.4299 0.5069 4.9856 0.1868 0.1428 0.6399 0.1968 0.0553 0.0261 0.0236 11.0040 10.7104 2.9352 1.6797 0.4982 0.1683 0.5174 0.1800 2.9935  Economic Valuation Of Critical Habitat Closures, Berman et al.  40  Black cod, average weight > 0.75 kg Limited Dependent Variable Model - CENSORED Ordinary least squares regression. Observations 263 Mean of LHS 0.1955044E+01 StdDev of resid. 0.1637353E+01 R-squared 0.5931938E+00 F[ 20, 242] Log-likelihood -0.4919181E+03 Amemiya Pr. Criter. 0.2894990E+01 ANOVA Source Regression Residual Total N(0,1) used for significance levels. Variable Coefficient Std. Error Constant 982.14 1653. J162 1.4588 0.7794 J177 2.9388 1.288 J192 3.2753 1.809 LDEPTH 1.1602 4.030 L2DEP 0.36078 0.3653 LGTEM 1.4113 1.407 LSLOPE 0.85200E-01 0.1255 LMLD 2.9710 2.025 LMTEM -9.6338 3.446 LMSTEM 6.5858 5.942 LMSAL -594.38 961.2 LMSSAL 30.364 11.99 VERTVEL -6.7201 4.589 HORVELM -0.96752E-01 0.4494E-01 MSHORVEL 0.71843E-01 0.4711E-01 SSHRE 0.29249 0.1014 LCHLA 0.69575 0.3049 LWIND -1.3734 0.9021 L2MSAL 90.307 139.8 LMLD_DEP -0.76351 0.4338  regression Dep. Variable Weights Std.Dev of LHS Sum of squares Adj. R-squared 0.1764389E+02 Restr.(b=0) Log-l Akaike Info.Crit. Variation 0.9460386E+03 0.6487836E+03 0.1594822E+04 t-ratio 0.594 1.872 2.282 1.811 0.288 0.988 1.003 0.679 1.467 -2.795 1.108 -0.618 2.532 -1.464 -2.153 1.525 2.884 2.282 -1.523 0.646 -1.760  Prob. 0.5524 0.0612 0.0225 0.0701 0.7735 0.3234 0.3159 0.4973 0.1423 0.0052 0.2677 0.5363 0.0113 0.1431 0.0313 0.1273 0.0039 0.0225 0.1279 0.5184 0.0784  LBBCOD ONE 0.2467206E+01 0.6487836E+03 0.5595734E+00 Prob value -0.6101916E+03 0.3900518E+01 Deg. freedom 20. 242. 262.  0.1147677E-35 Mean Square 0.4730193E+02 0.2680924E+01 0.6087107E+01  Var. Mean  Var. st. dev.  0.2776 0.2510 0.2433 4.8978 24.2447 1.7545 3.2086 1.5058 2.0028 -0.0900 3.4665 0.0127 -0.0032 15.3185 12.8499 -3.4664 0.7104 1.7980 12.0172 7.2726  0.4487 0.4344 0.4299 0.5069 4.9856 0.1428 0.9778 0.6399 0.1968 0.0553 0.0261 0.0195 0.0232 11.0040 10.7104 1.6797 0.4982 0.1683 0.1800 2.9935  ***************************************************************************** Limited Dependent Variable Model - CENSORED regression Maximum Likelihood Estimates Log-Likelihood.............. -289.92 Threshold values for the model: Lower 0.0000 Upper ********** N(0,1) used for significance levels. Variable Coefficient Std. Error t-ratio Prob. Var. Mean Constant 4045.1 3639. 1.112 0.2663 J162 3.7377 1.571 2.379 0.0174 0.2776 J177 8.5246 2.728 3.125 0.0018 0.2510 J192 8.7241 3.771 2.313 0.0207 0.2433 LDEPTH 67.698 11.49 5.892 0.0000 4.8978 L2DEP -5.3179 1.044 -5.091 0.0000 24.2447 LGTEM 12.842 3.608 3.560 0.0004 1.7545 LSLOPE 0.31280 0.2515 1.244 0.2137 3.2086 LMLD 7.5796 5.144 1.473 0.1406 1.5058 LMTEM -25.431 6.902 -3.685 0.0002 2.0028 LMSTEM 34.202 12.47 2.744 0.0061 -0.0900 LMSAL -2525.1 2113. -1.195 0.2322 3.4665 LMSSAL 66.766 20.31 3.288 0.0010 0.0127 VERTVEL -12.226 7.999 -1.529 0.1264 -0.0032 HORVELM -0.17359 0.8271E-01 -2.099 0.0358 15.3185 MSHORVEL 0.12531 0.8490E-01 1.476 0.1399 12.8499 SSHRE 0.79853 0.1900 4.204 0.0000 -3.4664 LCHLA 1.4385 0.5722 2.514 0.0119 0.7104 LWIND -7.3941 2.844 -2.600 0.0093 1.7980 L2MSAL 378.44 307.1 1.232 0.2179 12.0172 LMLD_DEP -2.0157 1.010 -1.995 0.0461 7.2726 Sigma 2.2199 0.1569 14.152 0.0000  Var. st. dev. 0.4487 0.4344 0.4299 0.5069 4.9856 0.1428 0.9778 0.6399 0.1968 0.0553 0.0261 0.0195 0.0232 11.0040 10.7104 1.6797 0.4982 0.1683 0.1800 2.9935  Economic Valuation Of Critical Habitat Closures, Berman et al.  41  Halibut, all Limited Dependent Variable Model - CENSORED regression Ordinary least squares regression. Dep. Variable LHAL Observations 263 Weights ONE Mean of LHS 0.3605741E+01 Std.Dev of LHS 0.2128749E+01 StdDev of resid. 0.1730145E+01 Sum of squares 0.7303901E+03 R-squared 0.3848164E+00 Adj. R-squared 0.3394340E+00 F[ 18, 244] 0.8479421E+01 Prob value Log-likelihood -0.5074982E+03 Restr.(b=0) Log-l -0.5713854E+03 Amemiya Pr. Criter. 0.3209655E+01 Akaike Info.Crit. 0.4003788E+01 ANOVA Source Variation Deg. freedom Regression 0.4568817E+03 18. Residual 0.7303901E+03 244. Total 0.1187272E+04 262. N(0,1) used for significance levels. Variable Coefficient Std. Error t-ratio Prob. Var. Mean Constant 44.989 14.16 3.177 0.0015 J162 0.75663 0.8639 0.876 0.3811 0.2776 J177 0.33069 1.355 0.244 0.8072 0.2510 J192 1.5001 1.870 0.802 0.4224 0.2433 L2DEP -0.37442 0.1345 -2.784 0.0054 24.2447 TIMELDEP -0.75653E-02 0.5589E-02 -1.354 0.1759 864.3330 LSTEM -1.6490 1.288 -1.280 0.2006 2.3153 LGTEM -35.150 14.24 -2.468 0.0136 1.7545 LSLOPE 0.14877 0.1247 1.193 0.2328 3.2086 LMLD -6.5972 1.900 -3.473 0.0005 1.5058 LMTEM 4.6740 2.798 1.671 0.0948 2.0028 LMSTEM -8.2449 6.496 -1.269 0.2043 -0.0900 LMSSAL -23.534 11.02 -2.135 0.0328 0.0127 VERTVEL -8.3319 4.776 -1.745 0.0811 -0.0032 LCHLA1 0.92643 0.5006 1.851 0.0642 0.8483 LCHLA2 -0.59286 0.4783 -1.239 0.2152 0.7532 LWIND -1.0785 0.9022 -1.195 0.2319 1.7980 L2GTEM 9.2649 3.838 2.414 0.0158 3.0986 LMLD_DEP 1.5925 0.4067 3.916 0.0001 7.2726 ***************************************************************************** Limited Dependent Variable Model - CENSORED regression Maximum Likelihood Estimates Log-Likelihood.............. -494.55 Threshold values for the model: Lower 0.0000 Upper ********** N(0,1) used for significance levels. Variable Coefficient Std. Error t-ratio Prob. Var. Mean Constant 54.944 17.38 3.162 0.0016 J162 1.0732 1.055 1.017 0.3090 0.2776 J177 0.45641 1.651 0.276 0.7822 0.2510 J192 1.9509 2.286 0.853 0.3935 0.2433 L2DEP -0.53155 0.1676 -3.172 0.0015 24.2447 TIMELDEP -0.93874E-02 0.6829E-02 -1.375 0.1693 864.3330 LSTEM -2.2799 1.565 -1.456 0.1453 2.3153 LGTEM -42.553 17.52 -2.429 0.0151 1.7545 LSLOPE 0.18518 0.1520 1.218 0.2232 3.2086 LMLD -9.7177 2.423 -4.010 0.0001 1.5058 LMTEM 6.4038 3.458 1.852 0.0640 2.0028 LMSTEM -10.717 7.975 -1.344 0.1790 -0.0900 LMSSAL -31.531 14.04 -2.247 0.0247 0.0127 VERTVEL -9.1408 5.735 -1.594 0.1110 -0.0032 LCHLA1 1.2576 0.6193 2.031 0.0423 0.8483 LCHLA2 -0.85596 0.5911 -1.448 0.1476 0.7532 LWIND -1.4691 1.079 -1.361 0.1735 1.7980 L2GTEM 11.049 4.707 2.347 0.0189 3.0986 LMLD_DEP 2.3412 0.5258 4.452 0.0000 7.2726 Sigma 2.0556 0.1081 19.013 0.0000  0.1578245E-16 Mean Square 0.2538232E+02 0.2993402E+01 0.4531572E+01 Var. st. dev. 0.4487 0.4344 0.4299 4.9856 121.2768 0.1868 0.1428 0.9778 0.6399 0.1968 0.0553 0.0195 0.0232 0.4133 0.4622 0.1683 0.5174 2.9935  Var. st. dev. 0.4487 0.4344 0.4299 4.9856 121.2768 0.1868 0.1428 0.9778 0.6399 0.1968 0.0553 0.0195 0.0232 0.4133 0.4622 0.1683 0.5174 2.9935  Economic Valuation Of Critical Habitat Closures, Berman et al.  42  Halibut, average weight > 1 kg Limited Dependent Variable Model - CENSORED Ordinary least squares regression. Observations 263 Mean of LHS 0.3518902E+01 StdDev of resid. 0.1869869E+01 R-squared 0.3191008E+00 F[ 18, 244] Log-likelihood -0.5279236E+03 Amemiya Pr. Criter. 0.3749004E+01 ANOVA Source Regression Residual Total N(0,1) used for significance levels. Variable Coefficient Std. Error Constant 11.477 26.56 J162 2.0763 0.7037 J177 2.2745 0.7962 J192 3.8644 1.145 LDEPTH 7.9750 6.291 L2DEP -1.3080 0.6029 LSTEM -2.4061 1.214 LGTEM -45.552 20.23 LSLOPE 0.14587 0.1339 LMLD -6.8059 2.245 LMSAL 8.8554 7.039 LMSSAL -15.872 12.68 VERTVEL -8.6660 5.158 LCHLA -0.46292 0.4457 LCHLA1 1.5084 0.7001 LCHLA2 -0.99361 0.5437 LWIND -0.90053 0.9947 L2GTEM 11.884 5.566 LMLD_DEP 1.6428 0.4881  regression Dep. Variable Weights Std.Dev of LHS Sum of squares Adj. R-squared 0.6352760E+01 Restr.(b=0) Log-l Akaike Info.Crit. Variation 0.3998135E+03 0.8531244E+03 0.1252938E+04 t-ratio 0.432 2.951 2.857 3.374 1.268 -2.169 -1.982 -2.252 1.089 -3.031 1.258 -1.252 -1.680 -1.039 2.155 -1.827 -0.905 2.135 3.366  Prob. 0.6657 0.0032 0.0043 0.0007 0.2049 0.0301 0.0475 0.0244 0.2760 0.0024 0.2084 0.2106 0.0929 0.2990 0.0312 0.0676 0.3653 0.0327 0.0008  LBHAL ONE 0.2186825E+01 0.8531244E+03 0.2688706E+00 Prob value -0.5784645E+03 0.4159115E+01 Deg. freedom 18. 244. 262.  0.8742407E-12 Mean Square 0.2221186E+02 0.3496411E+01 0.4782206E+01  Var. Mean  Var. st. dev.  0.2776 0.2510 0.2433 4.8978 24.2447 2.3153 1.7545 3.2086 1.5058 3.4665 0.0127 -0.0032 0.7104 0.8483 0.7532 1.7980 3.0986 7.2726  0.4487 0.4344 0.4299 0.5069 4.9856 0.1868 0.1428 0.9778 0.6399 0.0261 0.0195 0.0232 0.4982 0.4133 0.4622 0.1683 0.5174 2.9935  ***************************************************************************** Limited Dependent Variable Model - CENSORED regression Maximum Likelihood Estimates Log-Likelihood.............. -507.60 Threshold values for the model: Lower 0.0000 Upper ********** N(0,1) used for significance levels. Variable Coefficient Std. Error t-ratio Prob. Var. Mean Constant 10.845 33.49 0.324 0.7460 J162 2.9686 0.8959 3.314 0.0009 0.2776 J177 3.2217 1.005 3.205 0.0014 0.2510 J192 5.4084 1.453 3.722 0.0002 0.2433 LDEPTH 10.045 7.874 1.276 0.2021 4.8978 L2DEP -1.7466 0.7566 -2.309 0.0210 24.2447 LSTEM -3.2524 1.516 -2.146 0.0319 2.3153 LGTEM -57.220 25.20 -2.271 0.0232 1.7545 LSLOPE 0.18943 0.1675 1.131 0.2581 3.2086 LMLD -10.747 2.949 -3.644 0.0003 1.5058 LMSAL 12.680 9.100 1.393 0.1635 3.4665 LMSSAL -22.188 16.42 -1.351 0.1766 0.0127 VERTVEL -9.7879 6.355 -1.540 0.1235 -0.0032 LCHLA -0.68626 0.5705 -1.203 0.2290 0.7104 LCHLA1 2.1564 0.8975 2.403 0.0163 0.8483 LCHLA2 -1.4602 0.6995 -2.087 0.0369 0.7532 LWIND -1.3376 1.243 -1.076 0.2819 1.7980 L2GTEM 14.752 6.915 2.133 0.0329 3.0986 LMLD_DEP 2.5799 0.6495 3.972 0.0001 7.2726 Sigma 2.2760 0.1224 18.602 0.0000  Var. st. dev. 0.4487 0.4344 0.4299 0.5069 4.9856 0.1868 0.1428 0.9778 0.6399 0.0261 0.0195 0.0232 0.4982 0.4133 0.4622 0.1683 0.5174 2.9935  Economic Valuation Of Critical Habitat Closures, Berman et al.  43  Flatfish, all Limited Dependent Variable Model - CENSORED regression Ordinary least squares regression. Dep. Variable LFLAT Observations 263 Weights ONE Mean of LHS 0.5386837E+01 Std.Dev of LHS 0.1870004E+01 StdDev of resid. 0.1406774E+01 Sum of squares 0.4789211E+03 R-squared 0.4772697E+00 Adj. R-squared 0.4340689E+00 F[ 20, 242] 0.1104769E+02 Prob value Log-likelihood -0.4519995E+03 Restr.(b=0) Log-l -0.5373022E+03 Amemiya Pr. Criter. 0.2137033E+01 Akaike Info.Crit. 0.3596955E+01 ANOVA Source Variation Deg. freedom Regression 0.4372705E+03 20. Residual 0.4789211E+03 242. Total 0.9161916E+03 262. N(0,1) used for significance levels. Variable Coefficient Std. Error t-ratio Prob. Var. Mean Constant -2358.9 1289. -1.830 0.0672 J162 0.66428 0.5854 1.135 0.2565 0.2776 J177 -0.42864 1.012 -0.424 0.6719 0.2510 J192 -1.4083 1.426 -0.988 0.3232 0.2433 LDEPTH 32.631 4.354 7.495 0.0000 4.8978 L2DEP -2.8540 0.4220 -6.763 0.0000 24.2447 TIMELDEP -0.62780E-02 0.4152E-02 -1.512 0.1306 864.3330 LGTEM -16.487 15.51 -1.063 0.2878 1.7545 LSLOPE 0.33820 0.1084 3.119 0.0018 3.2086 LMTEM 4.7554 2.765 1.720 0.0854 2.0028 LMSAL 1384.7 747.8 1.852 0.0641 3.4665 LBMSAL -58.652 10.65 -5.509 0.0000 0.0380 VERTVEL 4.4290 3.923 1.129 0.2590 -0.0032 HORVELM -0.10453 0.3921E-01 -2.666 0.0077 15.3185 MSHORVEL 0.91852E-01 0.4088E-01 2.247 0.0246 12.8499 BMHORVEL -0.17495 0.4726E-01 -3.702 0.0002 2.8098 LCHLA1 0.48358 0.4142 1.167 0.2430 0.8483 LCHLA2 -0.76722 0.4142 -1.852 0.0640 0.7532 LWIND 1.6687 0.7756 2.151 0.0314 1.7980 L2GTEM 5.6575 4.273 1.324 0.1855 3.0986 L2MSAL -209.67 108.5 -1.932 0.0534 12.0172 ***************************************************************************** Limited Dependent Variable Model - CENSORED regression Maximum Likelihood Estimates Log-Likelihood.............. -455.13 Threshold values for the model: Lower 0.0000 Upper ********** N(0,1) used for significance levels. Variable Coefficient Std. Error t-ratio Prob. Var. Mean Constant -2443.5 1278. -1.913 0.0558 J162 0.52961 0.5841 0.907 0.3645 0.2776 J177 -0.84760 1.020 -0.831 0.4060 0.2510 J192 -1.9431 1.432 -1.357 0.1749 0.2433 LDEPTH 35.563 4.490 7.920 0.0000 4.8978 L2DEP -3.1205 0.4350 -7.173 0.0000 24.2447 TIMELDEP -0.67135E-02 0.4128E-02 -1.626 0.1039 864.3330 LGTEM -19.705 15.49 -1.272 0.2032 1.7545 LSLOPE 0.33531 0.1075 3.119 0.0018 3.2086 LMTEM 5.8421 2.774 2.106 0.0352 2.0028 LMSAL 1434.7 741.2 1.935 0.0529 3.4665 LBMSAL -62.703 10.64 -5.892 0.0000 0.0380 VERTVEL 4.6344 3.903 1.187 0.2351 -0.0032 HORVELM -0.96556E-01 0.3916E-01 -2.465 0.0137 15.3185 MSHORVEL 0.85591E-01 0.4072E-01 2.102 0.0355 12.8499 BMHORVEL -0.18795 0.4721E-01 -3.981 0.0001 2.8098 LCHLA1 0.46835 0.4126 1.135 0.2563 0.8483 LCHLA2 -0.76605 0.4146 -1.848 0.0646 0.7532 LWIND 1.6930 0.7700 2.199 0.0279 1.7980 L2GTEM 6.6162 4.270 1.549 0.1213 3.0986 L2MSAL -217.59 107.6 -2.022 0.0431 12.0172 Sigma 1.3934 0.6260E-01 22.258 0.0000  0.2508541E-23 Mean Square 0.2186353E+02 0.1979013E+01 0.3496915E+01 Var. st. dev. 0.4487 0.4344 0.4299 0.5069 4.9856 121.2768 0.1428 0.9778 0.1968 0.0261 0.0236 0.0232 11.0040 10.7104 2.9352 0.4133 0.4622 0.1683 0.5174 0.1800  Var. st. dev. 0.4487 0.4344 0.4299 0.5069 4.9856 121.2768 0.1428 0.9778 0.1968 0.0261 0.0236 0.0232 11.0040 10.7104 2.9352 0.4133 0.4622 0.1683 0.5174 0.1800  Economic Valuation Of Critical Habitat Closures, Berman et al.  44  Rockfish, all Ordinary Observations Mean of LHS StdDev of resid. R-squared F[ 22, Log-likelihood Amemiya Pr. Criter. ANOVA  least squares regression. Dep. Variable LROCK 263 Weights ONE 0.3033271E+01 Std.Dev of LHS 0.2706313E+01 0.2062773E+01 Sum of squares 0.1021207E+04 0.4678222E+00 Adj. R-squared 0.4190393E+00 240] 0.9589869E+01 Prob value -0.5515720E+03 Restr.(b=0) Log-l -0.6345193E+03 0.4627143E+01 Akaike Info.Crit. 0.4369369E+01 Source Variation Deg. freedom Regression 0.8977141E+03 22. Residual 0.1021207E+04 240. Total 0.1918921E+04 262. Variable Coefficient Std. Error t-ratio Prob. Var. Mean Constant 1504.2 1944. 0.774 0.4390 J162 -0.75671 0.8060 -0.939 0.3478 0.2776 J177 -0.27890 1.490 -0.187 0.8515 0.2510 J192 -0.76191 2.095 -0.364 0.7162 0.2433 LDEPTH -2.5107 6.417 -0.391 0.6956 4.8978 L2DEP 0.28820 0.6328 0.455 0.6488 24.2447 TIMELDEP 0.78801E-02 0.6279E-02 1.255 0.2095 864.3330 LGTEM 22.477 22.86 0.983 0.3254 1.7545 LSLOPE -0.70144E-01 0.1629 -0.430 0.6669 3.2086 LMTEM -6.4377 5.463 -1.178 0.2387 2.0028 LBMTEM -6.5865 2.821 -2.335 0.0196 -0.1522 LMSAL -916.79 1128. -0.813 0.4163 3.4665 LMSSAL 24.445 15.53 1.574 0.1154 0.0127 LBMSAL 29.658 15.83 1.874 0.0610 0.0380 VERTVEL -7.7301 5.687 -1.359 0.1740 -0.0032 HORVELM -0.32319E-01 0.1793E-01 -1.802 0.0715 15.3185 BMHORVEL 0.94278E-01 0.7277E-01 1.296 0.1951 2.8098 SSHRE -0.61026E-01 0.1284 -0.475 0.6345 -3.4664 LCHLA 1.4035 0.5139 2.731 0.0063 0.7104 LCHLA1 -1.0980 0.4983 -2.203 0.0276 0.8483 LWIND 0.34816 1.214 0.287 0.7742 1.7980 L2GTEM -6.3082 6.300 -1.001 0.3167 3.0986 L2MSAL 138.63 163.6 0.847 0.3969 12.0172 ***************************************************************************** Limited Dependent Variable Model - CENSORED regression Maximum Likelihood Estimates Log-Likelihood.............. -479.26 Threshold values for the model: Lower 0.0000 Upper ********** N(0,1) used for significance levels. Variable Coefficient Std. Error t-ratio Prob. Var. Mean Constant 2518.5 2439. 1.032 0.3019 J162 -0.94423 1.057 -0.894 0.3715 0.2776 J177 -0.23051 2.011 -0.115 0.9087 0.2510 J192 -0.70044 2.845 -0.246 0.8056 0.2433 LDEPTH 16.379 12.49 1.311 0.1899 4.8978 L2DEP -1.5830 1.226 -1.291 0.1967 24.2447 TIMELDEP 0.82648E-02 0.8136E-02 1.016 0.3097 864.3330 LGTEM 52.069 41.42 1.257 0.2088 1.7545 LSLOPE -0.22848 0.2106 -1.085 0.2781 3.2086 LMTEM -8.2066 6.995 -1.173 0.2407 2.0028 LBMTEM -11.663 3.758 -3.104 0.0019 -0.1522 LMSAL -1560.6 1416. -1.102 0.2705 3.4665 LMSSAL 24.014 19.63 1.224 0.2211 0.0127 LBMSAL 49.506 22.24 2.226 0.0260 0.0380 VERTVEL -8.9489 7.202 -1.243 0.2140 -0.0032 HORVELM -0.41086E-01 0.2331E-01 -1.763 0.0779 15.3185 BMHORVEL 0.19113 0.9863E-01 1.938 0.0527 2.8098 SSHRE -0.16915 0.1654 -1.023 0.3064 -3.4664 LCHLA 1.6080 0.6656 2.416 0.0157 0.7104 LCHLA1 -1.3805 0.6614 -2.087 0.0369 0.8483 LWIND 2.5312 2.195 1.153 0.2487 1.7980 L2GTEM -15.351 11.72 -1.310 0.1902 3.0986 L2MSAL 233.79 205.7 1.136 0.2558 12.0172  Sigma  2.4862  0.1358  18.308  0.0000  0.1974883E-21 Mean Square 0.4080519E+02 0.4255031E+01 0.7324128E+01 Var. st. dev. 0.4487 0.4344 0.4299 0.5069 4.9856 121.2768 0.1428 0.9778 0.1968 0.1850 0.0261 0.0195 0.0236 0.0232 11.0040 2.9352 1.6797 0.4982 0.4133 0.1683 0.5174 0.1800  Var. st. dev. 0.4487 0.4344 0.4299 0.5069 4.9856 121.2768 0.1428 0.9778 0.1968 0.1850 0.0261 0.0195 0.0236 0.0232 11.0040 2.9352 1.6797 0.4982 0.4133 0.1683 0.5174 0.1800  Economic Valuation Of Critical Habitat Closures, Berman et al.  45  2. Equations without wind Pacific cod, all Limited Dependent Variable Model - CENSORED regression Ordinary least squares regression. Dep. Variable LPCOD Observations 377 Weights ONE Mean of LHS 0.1951593E+01 Std.Dev of LHS 0.2121660E+01 StdDev of resid. 0.1870424E+01 Sum of squares 0.1259456E+04 R-squared 0.2558793E+00 Adj. R-squared 0.2228073E+00 F[ 16, 360] 0.7737031E+01 Prob value Log-likelihood -0.7623066E+03 Restr.(b=0) Log-l -0.8180181E+03 Amemiya Pr. Criter. 0.3656244E+01 Akaike Info.Crit. 0.4134252E+01 ANOVA Source Variation Deg. freedom Regression 0.4330865E+03 16. Residual 0.1259456E+04 360. Total 0.1692542E+04 376. N(0,1) used for significance levels. Variable Coefficient Std. Error t-ratio Prob. Var. Mean Constant 1423.6 690.7 2.061 0.0393 J162 0.52740 0.3813 1.383 0.1666 0.3263 J177 0.55088 0.4990 1.104 0.2696 0.2334 J192 0.90287 0.5848 1.544 0.1226 0.2334 LDEPTH 12.498 3.273 3.818 0.0001 4.8044 L2DEP -1.1314 0.3373 -3.354 0.0008 23.3976 LSTEM -2.6242 0.8848 -2.966 0.0030 2.2955 LGTEM 46.412 11.78 3.941 0.0001 1.7873 LSLOPE 0.43460 0.3255 1.335 0.1818 3.2026 L2SLOPE -0.66421E-01 0.6250E-01 -1.063 0.2879 11.2148 LMSAL -846.40 402.8 -2.101 0.0356 3.4545 LMSSAL -19.928 8.351 -2.386 0.0170 0.0135 LBMSAL -28.354 7.911 -3.584 0.0003 0.0410 MSHORVEL -0.95601E-02 0.1150E-01 -0.831 0.4060 10.9195 LCHLA2 0.47311 0.2280 2.075 0.0380 0.8487 L2GTEM -11.400 3.124 -3.649 0.0003 3.2248 L2MSAL 119.69 58.60 2.042 0.0411 11.9349 ***************************************************************************** Limited Dependent Variable Model - CENSORED regression Maximum Likelihood Estimates Log-Likelihood.............. -614.09 Threshold values for the model: Lower 0.0000 Upper ********** N(0,1) used for significance levels. Variable Coefficient Std. Error t-ratio Prob. Var. Mean Constant 2375.4 1188. 2.000 0.0455 J162 0.37777 0.6423 0.588 0.5564 0.3263 J177 -0.32678E-01 0.8509 -0.038 0.9694 0.2334 J192 0.46898 1.010 0.464 0.6424 0.2334 LDEPTH 33.600 6.485 5.181 0.0000 4.8044 L2DEP -3.2230 0.6741 -4.781 0.0000 23.3976 LSTEM -4.5960 1.464 -3.140 0.0017 2.2955 LGTEM 107.99 22.11 4.885 0.0000 1.7873 LSLOPE 1.5487 0.7957 1.946 0.0516 3.2026 L2SLOPE -0.25802 0.1450 -1.780 0.0751 11.2148 LMSAL -1448.1 692.3 -2.092 0.0365 3.4545 LMSSAL -35.562 14.72 -2.416 0.0157 0.0135 LBMSAL -45.527 12.99 -3.504 0.0005 0.0410 MSHORVEL -0.29942E-01 0.2054E-01 -1.458 0.1449 10.9195 LCHLA2 0.88429 0.3869 2.286 0.0223 0.8487 L2GTEM -26.392 5.780 -4.566 0.0000 3.2248 L2MSAL 204.76 100.7 2.034 0.0420 11.9349 Sigma 2.7979 0.1501 18.634 0.0000  0.8994911E-15 Mean Square 0.2706791E+02 0.3498488E+01 0.4501442E+01 Var. st. dev. 0.4695 0.4236 0.4236 0.5626 5.3509 0.1867 0.1745 0.9801 5.3014 0.0358 0.0274 0.0305 10.0521 0.5049 0.6559 0.2450  Var. st. dev. 0.4695 0.4236 0.4236 0.5626 5.3509 0.1867 0.1745 0.9801 5.3014 0.0358 0.0274 0.0305 10.0521 0.5049 0.6559 0.2450  46  Economic Valuation Of Critical Habitat Closures, Berman et al.  Pacific cod, average weight > 0.5 kg Limited Dependent Variable Model - CENSORED regression Ordinary least squares regression. Dep. Variable LBPCOD Observations 381 Weights ONE Mean of LHS 0.1749955E+01 Std.Dev of LHS 0.2159548E+01 StdDev of resid. 0.1960828E+01 Sum of squares 0.1391835E+04 R-squared 0.2146230E+00 Adj. R-squared 0.1755711E+00 F[ 18, 362] 0.5495840E+01 Prob value Log-likelihood -0.7874233E+03 Restr.(b=0) Log-l -0.8334465E+03 Amemiya Pr. Criter. 0.4036584E+01 Akaike Info.Crit. 0.4233193E+01 ANOVA Source Variation Deg. freedom Regression 0.3803519E+03 18. Residual 0.1391835E+04 362. Total 0.1772186E+04 380. N(0,1) used for significance levels. Variable Coefficient Std. Error t-ratio Prob. Var. Mean Constant 224.30 555.1 0.404 0.6862 J162 1.5314 0.6262 2.445 0.0145 0.3228 J177 1.6346 0.9857 1.658 0.0972 0.2336 J192 1.8798 1.395 1.347 0.1778 0.2257 LDEPTH 11.891 3.224 3.688 0.0002 4.7937 L2DEP -1.0322 0.3354 -3.077 0.0021 23.3024 LSTEM -1.7338 0.9764 -1.776 0.0758 2.2923 LGTEM 46.860 12.18 3.848 0.0001 1.7847 LSLOPE 0.42390 0.3414 1.242 0.2144 3.2009 L2SLOPE -0.56517E-01 0.6615E-01 -0.854 0.3929 11.1949 LMTEM -1.4291 2.297 -0.622 0.5338 1.9673 LMSTEM 6.0243 3.853 1.564 0.1179 -0.0912 LMSAL -145.42 324.5 -0.448 0.6540 3.4544 LBMSAL -33.179 8.041 -4.126 0.0000 0.0403 HORVELM 0.53499E-01 0.4454E-01 1.201 0.2297 12.9840 MSHORVEL -0.64555E-01 0.4692E-01 -1.376 0.1689 10.7954 LCHLA 0.39234 0.2179 1.801 0.0718 0.7226 L2GTEM -11.733 3.220 -3.644 0.0003 3.2155 L2MSAL 17.370 47.43 0.366 0.7142 11.9342 ***************************************************************************** Limited Dependent Variable Model - CENSORED regression Maximum Likelihood Estimates Log-Likelihood.............. -570.80 Threshold values for the model: Lower 0.0000 Upper ********** N(0,1) used for significance levels. Variable Coefficient Std. Error t-ratio Prob. Var. Mean Constant 1189.3 1114. 1.067 0.2858 J162 3.5561 1.311 2.713 0.0067 0.3228 J177 4.8798 2.087 2.338 0.0194 0.2336 J192 5.9164 2.954 2.003 0.0452 0.2257 LDEPTH 32.248 7.843 4.112 0.0000 4.7937 L2DEP -2.8980 0.8139 -3.561 0.0004 23.3024 LSTEM -4.0984 1.988 -2.061 0.0393 2.2923 LGTEM 218.32 42.14 5.181 0.0000 1.7847 LSLOPE 1.4928 1.027 1.453 0.1463 3.2009 L2SLOPE -0.20927 0.1871 -1.119 0.2633 11.1949 LMTEM -9.4971 5.004 -1.898 0.0577 1.9673 LMSTEM 15.221 8.363 1.820 0.0687 -0.0912 LMSAL -805.20 651.8 -1.235 0.2167 3.4544 LBMSAL -62.086 16.83 -3.689 0.0002 0.0403 HORVELM 0.13319 0.9186E-01 1.450 0.1471 12.9840 MSHORVEL -0.16556 0.9713E-01 -1.705 0.0883 10.7954 LCHLA 0.47038 0.4492 1.047 0.2951 0.7226 L2GTEM -56.370 11.21 -5.028 0.0000 3.2155 L2MSAL 110.77 95.23 1.163 0.2448 11.9342 Sigma 3.4294 0.2084 16.455 0.0000  0.1895555E-10 Mean Square 0.2113066E+02 0.3844847E+01 0.4663649E+01 Var. st. dev. 0.4682 0.4237 0.4186 0.5688 5.3937 0.1872 0.1747 0.9755 5.2778 0.2118 0.0537 0.0352 0.0301 10.5078 10.0679 0.5474 0.6557 0.2413  Var. st. dev. 0.4682 0.4237 0.4186 0.5688 5.3937 0.1872 0.1747 0.9755 5.2778 0.2118 0.0537 0.0352 0.0301 10.5078 10.0679 0.5474 0.6557 0.2413  Economic Valuation Of Critical Habitat Closures, Berman et al.  47  Pollock, all Limited Dependent Variable Model - CENSORED Ordinary least squares regression. Observations 380 Mean of LHS 0.1580394E+01 StdDev of resid. 0.1726682E+01 R-squared 0.3280923E+00 F[ 17, 362] Log-likelihood -0.7375332E+03 Amemiya Pr. Criter. 0.3122657E+01 ANOVA Source Regression Residual Total N(0,1) used for significance levels. Variable Coefficient Std. Error Constant -1937.3 643.6 J162 0.92638 0.4346 J177 1.2384 0.5070 J192 0.76996 0.7230 LDEPTH 3.8528 0.6249 LGTEM 38.828 8.212 LSLOPE -0.82837E-01 0.1030 LMLD 2.5593 1.333 LMSAL 1139.0 375.3 LBMSAL -27.278 7.020 VERTVEL 5.4075 4.504 HORVELM -0.19863E-01 0.1096E-01 BMHORVEL 0.45647E-01 0.4401E-01 LCHLA 0.72940 0.2705 LCHLA1 -0.23812 0.2881 L2GTEM -9.7652 2.148 L2MSAL -171.94 54.64 LMLD_DEP -0.50035 0.3005  regression Dep. Variable Weights Std.Dev of LHS Sum of squares Adj. R-squared 0.1039791E+02 Restr.(b=0) Log-l Akaike Info.Crit. Variation 0.5270111E+03 0.1079278E+04 0.1606289E+04 t-ratio -3.010 2.131 2.443 1.065 6.166 4.728 -0.804 1.920 3.035 -3.886 1.201 -1.812 1.037 2.696 -0.826 -4.547 -3.147 -1.665  Prob. 0.0026 0.0330 0.0146 0.2869 0.0000 0.0000 0.4211 0.0548 0.0024 0.0001 0.2299 0.0700 0.2996 0.0070 0.4086 0.0000 0.0016 0.0960  LPOLL ONE 0.2058696E+01 0.1079278E+04 0.2965386E+00 Prob value -0.8130837E+03 0.3976490E+01 Deg. freedom 17. 362. 379.  0.0000000E+00 Mean Square 0.3100065E+02 0.2981431E+01 0.4238230E+01  Var. Mean  Var. st. dev.  0.3237 0.2342 0.2237 4.7950 1.7844 3.1984 1.5198 3.4550 0.0399 -0.0025 13.0181 2.7418 0.7227 0.9196 3.2144 11.9379 7.1724  0.4685 0.4241 0.4173 0.5690 0.1748 0.9756 0.6260 0.0336 0.0287 0.0202 10.5005 2.9262 0.5481 0.4705 0.6562 0.2306 2.8464  ***************************************************************************** Limited Dependent Variable Model - CENSORED regression Maximum Likelihood Estimates Log-Likelihood.............. -628.14 Threshold values for the model: Lower 0.0000 Upper ********** N(0,1) used for significance levels. Variable Coefficient Std. Error t-ratio Prob. Var. Mean Constant -3281.3 1050. -3.126 0.0018 J162 1.0767 0.6594 1.633 0.1025 0.3237 J177 1.6396 0.7861 2.086 0.0370 0.2342 J192 0.80336 1.106 0.727 0.4675 0.2237 LDEPTH 5.8535 0.9975 5.868 0.0000 4.7950 LGTEM 60.649 13.48 4.498 0.0000 1.7844 LSLOPE -0.16146 0.1544 -1.045 0.2958 3.1984 LMLD 4.2850 2.116 2.025 0.0428 1.5198 LMSAL 1919.7 611.2 3.141 0.0017 3.4550 LBMSAL -38.469 10.70 -3.596 0.0003 0.0399 VERTVEL 9.0343 6.867 1.316 0.1883 -0.0025 HORVELM -0.36579E-01 0.1690E-01 -2.164 0.0304 13.0181 BMHORVEL 0.12661 0.6929E-01 1.827 0.0677 2.7418 LCHLA 0.88612 0.4046 2.190 0.0285 0.7227 LCHLA1 -0.44773 0.4311 -1.038 0.2991 0.9196 L2GTEM -15.501 3.570 -4.342 0.0000 3.2144 L2MSAL -287.83 89.00 -3.234 0.0012 11.9379 LMLD_DEP -0.86254 0.4681 -1.843 0.0654 7.1724 Sigma 2.4187 0.1218 19.859 0.0000  Var. st. dev. 0.4685 0.4241 0.4173 0.5690 0.1748 0.9756 0.6260 0.0336 0.0287 0.0202 10.5005 2.9262 0.5481 0.4705 0.6562 0.2306 2.8464  48  Economic Valuation Of Critical Habitat Closures, Berman et al.  Pollock, average weight > 0.25 kg Limited Dependent Variable Model - CENSORED regression Ordinary least squares regression. Dep. Variable LBPOLL Observations 380 Weights ONE Mean of LHS 0.1134761E+01 Std.Dev of LHS 0.1956043E+01 StdDev of resid. 0.1648476E+01 Sum of squares 0.9918778E+03 R-squared 0.3159901E+00 Adj. R-squared 0.2897541E+00 F[ 14, 365] 0.1204414E+02 Prob value Log-likelihood -0.7214881E+03 Restr.(b=0) Log-l -0.7936468E+03 Amemiya Pr. Criter. 0.2824742E+01 Akaike Info.Crit. 0.3876253E+01 ANOVA Source Variation Deg. freedom Regression 0.4582150E+03 14. Residual 0.9918778E+03 365. Total 0.1450093E+04 379. N(0,1) used for significance levels. Variable Coefficient Std. Error t-ratio Prob. Var. Mean Constant -1348.8 592.5 -2.276 0.0228 J162 0.89908 0.4101 2.192 0.0284 0.3237 J177 1.1277 0.4548 2.480 0.0132 0.2342 J192 1.2104 0.6751 1.793 0.0730 0.2237 LDEPTH 3.9858 0.5955 6.693 0.0000 4.7950 LGTEM 34.701 7.729 4.490 0.0000 1.7844 LMLD 3.2231 1.268 2.542 0.0110 1.5198 LMSAL 797.98 345.8 2.308 0.0210 3.4550 LBMSAL -31.406 6.667 -4.711 0.0000 0.0399 HORVELM -0.11829E-01 0.9293E-02 -1.273 0.2031 13.0181 LCHLA 0.95612 0.2569 3.721 0.0002 0.7227 LCHLA1 -0.45142 0.2729 -1.654 0.0981 0.9196 L2GTEM -8.6805 2.021 -4.296 0.0000 3.2144 L2MSAL -122.34 50.37 -2.429 0.0151 11.9379 LMLD_DEP -0.59225 0.2860 -2.071 0.0384 7.1724 ***************************************************************************** Limited Dependent Variable Model - CENSORED regression Maximum Likelihood Estimates Log-Likelihood.............. -421.77 Threshold values for the model: Lower 0.0000 Upper ********** N(0,1) used for significance levels. Variable Coefficient Std. Error t-ratio Prob. Var. Mean Constant -4805.1 1799. -2.672 0.0075 J162 3.1049 1.175 2.643 0.0082 0.3237 J177 3.4288 1.305 2.628 0.0086 0.2342 J192 4.6284 1.930 2.399 0.0165 0.2237 LDEPTH 12.433 2.086 5.961 0.0000 4.7950 LGTEM 127.18 30.49 4.171 0.0000 1.7844 LMLD 12.360 4.333 2.853 0.0043 1.5198 LMSAL 2799.5 1046. 2.676 0.0075 3.4550 LBMSAL -82.426 18.95 -4.350 0.0000 0.0399 HORVELM -0.35186E-01 0.2606E-01 -1.350 0.1769 13.0181 LCHLA 2.7487 0.7121 3.860 0.0001 0.7227 LCHLA1 -1.6501 0.7357 -2.243 0.0249 0.9196 L2GTEM -33.342 8.316 -4.009 0.0001 3.2144 L2MSAL -423.15 152.5 -2.775 0.0055 11.9379 LMLD_DEP -2.1187 0.9060 -2.339 0.0194 7.1724 Sigma 3.5479 0.2618 13.551 0.0000  0.4875549E-22 Mean Square 0.3272964E+02 0.2717473E+01 0.3826102E+01 Var. st. dev. 0.4685 0.4241 0.4173 0.5690 0.1748 0.6260 0.0336 0.0287 10.5005 0.5481 0.4705 0.6562 0.2306 2.8464  Var. st. dev. 0.4685 0.4241 0.4173 0.5690 0.1748 0.6260 0.0336 0.0287 10.5005 0.5481 0.4705 0.6562 0.2306 2.8464  Economic Valuation Of Critical Habitat Closures, Berman et al.  49  Black cod, all Limited Dependent Variable Model - CENSORED regression Ordinary least squares regression. Dep. Variable LBCOD Observations 380 Weights ONE Mean of LHS 0.1691693E+01 Std.Dev of LHS 0.2331162E+01 StdDev of resid. 0.1611987E+01 Sum of squares 0.9458542E+03 R-squared 0.5407595E+00 Adj. R-squared 0.5218347E+00 F[ 15, 364] 0.2857420E+02 Prob value Log-likelihood -0.7124609E+03 Restr.(b=0) Log-l -0.8603153E+03 Amemiya Pr. Criter. 0.2707911E+01 Akaike Info.Crit. 0.3834005E+01 ANOVA Source Variation Deg. freedom Regression 0.1113751E+04 15. Residual 0.9458542E+03 364. Total 0.2059605E+04 379. N(0,1) used for significance levels. Variable Coefficient Std. Error t-ratio Prob. Var. Mean Constant -11.569 14.36 -0.806 0.4204 J162 1.5302 0.5078 3.013 0.0026 0.3237 J177 2.0762 0.8046 2.580 0.0099 0.2342 J192 3.0962 1.111 2.786 0.0053 0.2237 LDEPTH 2.6106 0.2619 9.967 0.0000 4.7950 LGTEM -3.8113 7.670 -0.497 0.6193 1.7844 LSLOPE 0.53129E-01 0.9596E-01 0.554 0.5798 3.1984 LMTEM -4.5406 1.910 -2.377 0.0175 1.9668 LMSTEM 10.651 3.081 3.457 0.0005 -0.0907 LBMTEM -4.4453 1.154 -3.851 0.0001 -0.1087 LMSAL 3.2199 4.153 0.775 0.4381 3.4550 VERTVEL -6.2213 4.241 -1.467 0.1424 -0.0025 HORVELM -0.62759E-01 0.3714E-01 -1.690 0.0911 13.0181 MSHORVEL 0.47594E-01 0.3944E-01 1.207 0.2275 10.8239 LCHLA1 0.49223 0.1937 2.541 0.0111 0.9196 L2GTEM 1.1889 2.007 0.592 0.5536 3.2144 ***************************************************************************** Limited Dependent Variable Model - CENSORED regression Maximum Likelihood Estimates Log-Likelihood.............. -427.20 Threshold values for the model: Lower 0.0000 Upper ********** N(0,1) used for significance levels. Variable Coefficient Std. Error t-ratio Prob. Var. Mean Constant -186.07 37.80 -4.923 0.0000 J162 3.5569 1.245 2.856 0.0043 0.3237 J177 6.3356 2.073 3.056 0.0022 0.2342 J192 9.0142 2.881 3.129 0.0018 0.2237 LDEPTH 6.9424 0.7007 9.908 0.0000 4.7950 LGTEM 132.99 27.97 4.754 0.0000 1.7844 LSLOPE 0.27057 0.2252 1.202 0.2295 3.1984 LMTEM -15.878 4.536 -3.500 0.0005 1.9668 LMSTEM 16.773 8.181 2.050 0.0403 -0.0907 LBMTEM -9.3727 2.637 -3.555 0.0004 -0.1087 LMSAL 16.582 9.264 1.790 0.0735 3.4550 VERTVEL -9.2226 8.656 -1.066 0.2866 -0.0025 HORVELM -0.19006 0.7697E-01 -2.469 0.0135 13.0181 MSHORVEL 0.15606 0.7938E-01 1.966 0.0493 10.8239 LCHLA1 1.2566 0.4294 2.927 0.0034 0.9196 L2GTEM -36.502 7.835 -4.659 0.0000 3.2144 Sigma 2.5864 0.1625 15.913 0.0000  0.0000000E+00 Mean Square 0.7425007E+02 0.2598500E+01 0.5434315E+01 Var. st. dev. 0.4685 0.4241 0.4173 0.5690 0.1748 0.9756 0.2118 0.0529 0.1885 0.0336 0.0202 10.5005 10.0659 0.4705 0.6562  Var. st. dev. 0.4685 0.4241 0.4173 0.5690 0.1748 0.9756 0.2118 0.0529 0.1885 0.0336 0.0202 10.5005 10.0659 0.4705 0.6562  50  Economic Valuation Of Critical Habitat Closures, Berman et al.  Black cod, average weight > 0.75 kg Limited Dependent Variable Model - CENSORED regression Ordinary least squares regression. Dep. Variable LBBCOD Observations 374 Weights ONE Mean of LHS 0.1696323E+01 Std.Dev of LHS 0.2349267E+01 StdDev of resid. 0.1628311E+01 Sum of squares 0.9438974E+03 R-squared 0.5414875E+00 Adj. R-squared 0.5195922E+00 F[ 17, 356] 0.2473081E+02 Prob value Log-likelihood -0.7038005E+03 Restr.(b=0) Log-l -0.8496170E+03 Amemiya Pr. Criter. 0.2779005E+01 Akaike Info.Crit. 0.3859895E+01 ANOVA Source Variation Deg. freedom Regression 0.1114710E+04 17. Residual 0.9438974E+03 356. Total 0.2058608E+04 373. N(0,1) used for significance levels. Variable Coefficient Std. Error t-ratio Prob. Var. Mean Constant -15.959 20.30 -0.786 0.4317 J162 1.7021 0.5732 2.969 0.0030 0.3289 J177 2.2423 0.8924 2.513 0.0120 0.2353 J192 3.2224 1.185 2.719 0.0065 0.2273 LDEPTH 2.9182 0.3463 8.427 0.0000 4.8091 LGTEM -10.021 8.246 -1.215 0.2242 1.7859 LSLOPE 0.11155 0.9676E-01 1.153 0.2490 3.1972 LMTEM -2.0865 1.928 -1.082 0.2792 1.9714 LMSTEM 8.5575 3.145 2.721 0.0065 -0.0915 LMSAL 4.6443 6.711 0.692 0.4889 3.4556 LMSSAL 9.1180 6.368 1.432 0.1522 0.0127 LBMSAL -9.0701 6.851 -1.324 0.1855 0.0404 HORVELM -0.56622E-01 0.3755E-01 -1.508 0.1316 13.1751 MSHORVEL 0.51108E-01 0.3954E-01 1.293 0.1962 10.9839 SSHRE 0.12401 0.5373E-01 2.308 0.0210 -3.5265 LCHLA1 0.83496 0.3605 2.316 0.0206 0.9176 LCHLA2 -0.43952 0.3528 -1.246 0.2128 0.8485 L2GTEM 2.6929 2.173 1.239 0.2153 3.2197 ***************************************************************************** Limited Dependent Variable Model - CENSORED regression Maximum Likelihood Estimates Log-Likelihood.............. -391.25 Threshold values for the model: Lower 0.0000 Upper ********** N(0,1) used for significance levels. Variable Coefficient Std. Error t-ratio Prob. Var. Mean Constant -466.94 89.95 -5.191 0.0000 J162 4.1629 1.509 2.759 0.0058 0.3289 J177 7.3562 2.529 2.909 0.0036 0.2353 J192 8.7336 3.343 2.613 0.0090 0.2273 LDEPTH 7.1558 0.9911 7.220 0.0000 4.8091 LGTEM 115.82 32.66 3.546 0.0004 1.7859 LSLOPE 0.31719 0.2308 1.374 0.1694 3.1972 LMTEM -10.210 5.835 -1.750 0.0801 1.9714 LMSTEM 13.796 8.799 1.568 0.1169 -0.0915 LMSAL 97.421 29.22 3.334 0.0009 3.4556 LMSSAL 34.048 15.62 2.180 0.0292 0.0127 LBMSAL 70.359 31.05 2.266 0.0234 0.0404 HORVELM -0.17467 0.8019E-01 -2.178 0.0294 13.1751 MSHORVEL 0.13521 0.7991E-01 1.692 0.0906 10.9839 SSHRE 0.17729 0.1218 1.455 0.1456 -3.5265 LCHLA1 2.1955 0.8847 2.482 0.0131 0.9176 LCHLA2 -1.1974 0.8400 -1.425 0.1540 0.8485 L2GTEM -31.303 9.360 -3.344 0.0008 3.2197 Sigma 2.4939 0.1595 15.633 0.0000  0.0000000E+00 Mean Square 0.6557120E+02 0.2651397E+01 0.5519056E+01 Var. st. dev. 0.4704 0.4248 0.4196 0.5586 0.1739 0.9813 0.2087 0.0528 0.0334 0.0241 0.0286 10.4944 10.0643 1.8770 0.4615 0.5044 0.6533  Var. st. dev. 0.4704 0.4248 0.4196 0.5586 0.1739 0.9813 0.2087 0.0528 0.0334 0.0241 0.0286 10.4944 10.0643 1.8770 0.4615 0.5044 0.6533  Economic Valuation Of Critical Habitat Closures, Berman et al.  51  Halibut, all Limited Dependent Variable Model - CENSORED Ordinary least squares regression. Observations 374 Mean of LHS 0.3650597E+01 StdDev of resid. 0.1690549E+01 R-squared 0.3756375E+00 F[ 16, 357] Log-likelihood -0.7183536E+03 Amemiya Pr. Criter. 0.2987862E+01 ANOVA Source Regression Residual Total N(0,1) used for significance levels. Variable Coefficient Std. Error Constant 9.9394 13.38 J162 1.4329 0.4986 J177 1.8114 0.7062 J192 3.0988 1.052 LDEPTH -3.5200 1.056 TIMELDEP -0.65529E-02 0.4476E-02 LGTEM -1.8108 0.9826 LSLOPE 0.26355 0.9636E-01 LMLD -6.3529 1.375 LMSAL 4.6533 3.664 LMSSAL -5.7320 6.048 VERTVEL -6.9483 4.442 SSHRE 0.91710E-01 0.5381E-01 LCHLA -0.44326 0.2856 LCHLA1 0.50492 0.4884 LCHLA2 -0.43661 0.3964 LMLD_DEP 1.5506 0.3099  regression Dep. Variable Weights Std.Dev of LHS Sum of squares Adj. R-squared 0.1342395E+02 Restr.(b=0) Log-l Akaike Info.Crit. Variation 0.6138407E+03 0.1020290E+04 0.1634131E+04 t-ratio 0.743 2.874 2.565 2.946 -3.332 -1.464 -1.843 2.735 -4.621 1.270 -0.948 -1.564 1.704 -1.552 1.034 -1.102 5.004  Prob. 0.4577 0.0041 0.0103 0.0032 0.0009 0.1432 0.0653 0.0062 0.0000 0.2041 0.3433 0.1177 0.0883 0.1206 0.3013 0.2707 0.0000  LHAL ONE 0.2093095E+01 0.1020290E+04 0.3476549E+00 Prob value -0.8064352E+03 0.3932372E+01 Deg. freedom 16. 357. 373.  0.4611915E-27 Mean Square 0.3836505E+02 0.2857955E+01 0.4381047E+01  Var. Mean  Var. st. dev.  0.3289 0.2353 0.2273 4.8091 846.4649 1.7859 3.1972 1.5095 3.4556 0.0127 -0.0025 -3.5265 0.7189 0.9176 0.8485 7.1526  0.4704 0.4248 0.4196 0.5586 124.6727 0.1739 0.9813 0.6250 0.0334 0.0241 0.0202 1.8770 0.5424 0.4615 0.5044 2.8620  ***************************************************************************** Limited Dependent Variable Model - CENSORED regression Maximum Likelihood Estimates Log-Likelihood.............. -704.05 Threshold values for the model: Lower 0.0000 Upper ********** N(0,1) used for significance levels. Variable Coefficient Std. Error t-ratio Prob. Var. Mean Constant 11.614 16.40 0.708 0.4788 J162 1.8384 0.6079 3.024 0.0025 0.3289 J177 2.3043 0.8580 2.686 0.0072 0.2353 J192 4.0739 1.286 3.167 0.0015 0.2273 LDEPTH -5.0082 1.320 -3.795 0.0001 4.8091 TIMELDEP -0.80978E-02 0.5475E-02 -1.479 0.1391 846.4649 LGTEM -2.6236 1.198 -2.189 0.0286 1.7859 LSLOPE 0.31328 0.1187 2.640 0.0083 3.1972 LMLD -9.5375 1.780 -5.358 0.0000 1.5095 LMSAL 6.6920 4.539 1.474 0.1404 3.4556 LMSSAL -12.453 8.046 -1.548 0.1217 0.0127 VERTVEL -7.7724 5.352 -1.452 0.1465 -0.0025 SSHRE 0.11360 0.6591E-01 1.723 0.0848 -3.5265 LCHLA -0.59920 0.3528 -1.698 0.0894 0.7189 LCHLA1 0.71807 0.5988 1.199 0.2304 0.9176 LCHLA2 -0.57077 0.4851 -1.177 0.2394 0.8485 LMLD_DEP 2.2913 0.4032 5.683 0.0000 7.1526 Sigma 2.0165 0.8828E-01 22.841 0.0000  Var. st. dev. 0.4704 0.4248 0.4196 0.5586 124.6727 0.1739 0.9813 0.6250 0.0334 0.0241 0.0202 1.8770 0.5424 0.4615 0.5044 2.8620  Economic Valuation Of Critical Habitat Closures, Berman et al.  52  Halibut, average weight > 1 kg Limited Dependent Variable Model - CENSORED Ordinary least squares regression. Observations 374 Mean of LHS 0.3444155E+01 StdDev of resid. 0.1933219E+01 R-squared 0.2779140E+00 F[ 19, 354] Log-likelihood -0.7669414E+03 Amemiya Pr. Criter. 0.3937192E+01 ANOVA Source Regression Residual Total N(0,1) used for significance levels. Variable Coefficient Std. Error Constant -17.851 16.40 J162 2.1568 0.5878 J177 2.9741 0.8362 J192 4.6804 1.250 LDEPTH 6.9992 3.233 L2DEP -1.0170 0.2741 TIMELDEP -0.85529E-02 0.5611E-02 LSTEM -1.2177 0.9641 LGTEM -2.2955 1.137 LSLOPE 0.20461 0.1130 LMLD -6.5637 1.674 LMSAL 6.1386 4.272 LMSSAL -5.4676 7.412 VERTVEL -6.1099 5.097 BMHORVEL 0.73806E-01 0.4734E-01 SSHRE 0.12446 0.6225E-01 LCHLA -0.29316 0.3297 LCHLA1 0.49847 0.5590 LCHLA2 -0.53682 0.4543 LMLD_DEP 1.6323 0.3694  regression Dep. Variable Weights Std.Dev of LHS Sum of squares Adj. R-squared 0.7170858E+01 Restr.(b=0) Log-l Akaike Info.Crit. Variation 0.5091980E+03 0.1323016E+04 0.1832214E+04 t-ratio -1.089 3.669 3.557 3.745 2.165 -3.710 -1.524 -1.263 -2.019 1.811 -3.921 1.437 -0.738 -1.199 1.559 1.999 -0.889 0.892 -1.182 4.419  Prob. 0.2763 0.0002 0.0004 0.0002 0.0304 0.0002 0.1274 0.2066 0.0435 0.0702 0.0001 0.1508 0.4607 0.2306 0.1190 0.0456 0.3739 0.3725 0.2374 0.0000  LBHAL ONE 0.2216326E+01 0.1323016E+04 0.2391580E+00 Prob value -0.8278307E+03 0.4208243E+01 Deg. freedom 19. 354. 373.  0.0000000E+00 Mean Square 0.2679990E+02 0.3737335E+01 0.4912103E+01  Var. Mean  Var. st. dev.  0.3289 0.2353 0.2273 4.8091 23.4383 846.4649 2.2942 1.7859 3.1972 1.5095 3.4556 0.0127 -0.0025 2.7209 -3.5265 0.7189 0.9176 0.8485 7.1526  0.4704 0.4248 0.4196 0.5586 5.3224 124.6727 0.1865 0.1739 0.9813 0.6250 0.0334 0.0241 0.0202 2.9310 1.8770 0.5424 0.4615 0.5044 2.8620  ***************************************************************************** Limited Dependent Variable Model - CENSORED regression Maximum Likelihood Estimates Log-Likelihood.............. -730.62 Threshold values for the model: Lower 0.0000 Upper ********** N(0,1) used for significance levels. Variable Coefficient Std. Error t-ratio Prob. Var. Mean Constant -26.168 21.25 -1.231 0.2182 J162 2.9904 0.7632 3.918 0.0001 0.3289 J177 4.0626 1.084 3.747 0.0002 0.2353 J192 6.5508 1.633 4.013 0.0001 0.2273 LDEPTH 9.6981 4.345 2.232 0.0256 4.8091 L2DEP -1.4655 0.3729 -3.930 0.0001 23.4383 TIMELDEP -0.11694E-01 0.7292E-02 -1.604 0.1088 846.4649 LSTEM -1.5266 1.247 -1.224 0.2209 2.2942 LGTEM -3.3208 1.476 -2.250 0.0244 1.7859 LSLOPE 0.24185 0.1470 1.646 0.0998 3.1972 LMLD -10.655 2.305 -4.624 0.0000 1.5095 LMSAL 8.7282 5.634 1.549 0.1213 3.4556 LMSSAL -13.145 10.58 -1.242 0.2141 0.0127 VERTVEL -7.1632 6.462 -1.108 0.2677 -0.0025 BMHORVEL 0.10196 0.6190E-01 1.647 0.0995 2.7209 SSHRE 0.16483 0.8050E-01 2.048 0.0406 -3.5265 LCHLA -0.44976 0.4303 -1.045 0.2959 0.7189 LCHLA1 0.81429 0.7266 1.121 0.2624 0.9176 LCHLA2 -0.77288 0.5905 -1.309 0.1906 0.8485 LMLD_DEP 2.6076 0.5114 5.099 0.0000 7.1526 Sigma 2.4188 0.1108 21.824 0.0000  Var. st. dev. 0.4704 0.4248 0.4196 0.5586 5.3224 124.6727 0.1865 0.1739 0.9813 0.6250 0.0334 0.0241 0.0202 2.9310 1.8770 0.5424 0.4615 0.5044 2.8620  Economic Valuation Of Critical Habitat Closures, Berman et al.  53  Flatfish, all (obervations where wind data available) Limited Dependent Variable Model - CENSORED regression Ordinary least squares regression. Dep. Variable LFLAT Observations 263 Weights ONE Mean of LHS 0.5386837E+01 Std.Dev of LHS 0.1870004E+01 StdDev of resid. 0.1417238E+01 Sum of squares 0.4880812E+03 R-squared 0.4672717E+00 Adj. R-squared 0.4256181E+00 F[ 19, 243] 0.1121803E+02 Prob value Log-likelihood -0.4544909E+03 Restr.(b=0) Log-l -0.5373022E+03 Amemiya Pr. Criter. 0.2161307E+01 Akaike Info.Crit. 0.3608296E+01 ANOVA Source Variation Deg. freedom Regression 0.4281105E+03 19. Residual 0.4880812E+03 243. Total 0.9161916E+03 262. N(0,1) used for significance levels. Variable Coefficient Std. Error t-ratio Prob. Var. Mean Constant -2553.6 1295. -1.972 0.0487 J162 0.77812 0.5873 1.325 0.1852 0.2776 J177 -0.23570 1.016 -0.232 0.8165 0.2510 J192 -1.4423 1.436 -1.004 0.3152 0.2433 LDEPTH 33.119 4.380 7.561 0.0000 4.8978 L2DEP -2.9114 0.4243 -6.862 0.0000 24.2447 TIMELDEP -0.66245E-02 0.4180E-02 -1.585 0.1130 864.3330 LGTEM -19.776 15.55 -1.272 0.2034 1.7545 LSLOPE 0.33226 0.1092 3.042 0.0023 3.2086 LMTEM 4.0266 2.764 1.457 0.1452 2.0028 LMSAL 1494.1 751.6 1.988 0.0468 3.4665 LBMSAL -52.429 10.32 -5.079 0.0000 0.0380 VERTVEL 3.9298 3.946 0.996 0.3193 -0.0032 HORVELM -0.10193 0.3948E-01 -2.582 0.0098 15.3185 MSHORVEL 0.88084E-01 0.4114E-01 2.141 0.0323 12.8499 BMHORVEL -0.16235 0.4724E-01 -3.437 0.0006 2.8098 LCHLA1 0.47578 0.4173 1.140 0.2542 0.8483 LCHLA2 -0.74886 0.4172 -1.795 0.0726 0.7532 L2GTEM 6.5509 4.285 1.529 0.1263 3.0986 L2MSAL -224.49 109.1 -2.057 0.0397 12.0172 ***************************************************************************** Limited Dependent Variable Model - CENSORED regression Maximum Likelihood Estimates Log-Likelihood.............. -457.53 Threshold values for the model: Lower 0.0000 Upper ********** N(0,1) used for significance levels. Variable Coefficient Std. Error t-ratio Prob. Var. Mean Constant -2642.8 1286. -2.055 0.0399 J162 0.63877 0.5875 1.087 0.2770 0.2776 J177 -0.66665 1.027 -0.649 0.5162 0.2510 J192 -1.9952 1.446 -1.380 0.1677 0.2433 LDEPTH 36.156 4.533 7.976 0.0000 4.8978 L2DEP -3.1888 0.4389 -7.266 0.0000 24.2447 TIMELDEP -0.70319E-02 0.4163E-02 -1.689 0.0912 864.3330 LGTEM -23.145 15.55 -1.488 0.1367 1.7545 LSLOPE 0.32942 0.1085 3.036 0.0024 3.2086 LMTEM 5.1148 2.779 1.840 0.0657 2.0028 LMSAL 1546.6 746.4 2.072 0.0383 3.4665 LBMSAL -56.455 10.35 -5.457 0.0000 0.0380 VERTVEL 4.0933 3.931 1.041 0.2977 -0.0032 HORVELM -0.93561E-01 0.3952E-01 -2.367 0.0179 15.3185 MSHORVEL 0.81467E-01 0.4106E-01 1.984 0.0473 12.8499 BMHORVEL -0.17543 0.4729E-01 -3.709 0.0002 2.8098 LCHLA1 0.45520 0.4164 1.093 0.2743 0.8483 LCHLA2 -0.74056 0.4183 -1.771 0.0766 0.7532 L2GTEM 7.5530 4.289 1.761 0.0782 3.0986 L2MSAL -232.77 108.4 -2.148 0.0317 12.0172 Sigma 1.4064 0.6318E-01 22.259 0.0000  0.0000000E+00 Mean Square 0.2253213E+02 0.2008564E+01 0.3496915E+01 Var. st. dev. 0.4487 0.4344 0.4299 0.5069 4.9856 121.2768 0.1428 0.9778 0.1968 0.0261 0.0236 0.0232 11.0040 10.7104 2.9352 0.4133 0.4622 0.5174 0.1800  Var. st. dev. 0.4487 0.4344 0.4299 0.5069 4.9856 121.2768 0.1428 0.9778 0.1968 0.0261 0.0236 0.0232 11.0040 10.7104 2.9352 0.4133 0.4622 0.5174 0.1800  Economic Valuation Of Critical Habitat Closures, Berman et al.  54  Flatfish, all, full sample Limited Dependent Variable Model - CENSORED Ordinary least squares regression. Observations 381 Mean of LHS 0.5595923E+01 StdDev of resid. 0.1528464E+01 R-squared 0.3107195E+00 F[ 16, 364] Log-likelihood -0.6935645E+03 Amemiya Pr. Criter. 0.2440443E+01 ANOVA Source Regression Residual Total N(0,1) used for significance levels. Variable Coefficient Std. Error Constant 66.658 16.34 J162 0.57483 0.3955 J177 0.47079 0.4818 J192 -0.32654 0.6573 LDEPTH 11.020 2.312 L2DEP -0.98788 0.2190 LGTEM 1.7617 0.9067 LSLOPE 0.30548 0.9090E-01 LMLD -1.7468 1.243 LMSTEM 6.0196 3.273 LMSAL -26.729 4.994 LBMSAL -22.951 5.972 HORVELM -0.11637 0.3563E-01 MSHORVEL 0.90229E-01 0.3723E-01 BMHORVEL -0.97203E-01 0.3809E-01 LCHLA -0.26639 0.1685 LMLD_DEP 0.31329 0.2699  regression Dep. Variable Weights Std.Dev of LHS Sum of squares Adj. R-squared 0.1025543E+02 Restr.(b=0) Log-l Akaike Info.Crit. Variation 0.3833403E+03 0.8503779E+03 0.1233718E+04 t-ratio 4.078 1.453 0.977 -0.497 4.767 -4.510 1.943 3.361 -1.405 1.839 -5.353 -3.843 -3.266 2.423 -2.552 -1.581 1.161  Prob. 0.0000 0.1462 0.3285 0.6193 0.0000 0.0000 0.0520 0.0008 0.1601 0.0659 0.0000 0.0001 0.0011 0.0154 0.0107 0.1139 0.2457  LFLAT ONE 0.1801840E+01 0.8503779E+03 0.2804214E+00 Prob value -0.7644509E+03 0.3729997E+01 Deg. freedom 16. 364. 380.  0.1805557E-20 Mean Square 0.2395877E+02 0.2336203E+01 0.3246627E+01  Var. Mean  Var. st. dev.  0.3228 0.2336 0.2257 4.7937 23.3024 1.7847 3.2009 1.5147 -0.0912 3.4544 0.0403 12.9840 10.7954 2.7609 0.7226 7.1489  0.4682 0.4237 0.4186 0.5688 5.3937 0.1747 0.9755 0.6330 0.0537 0.0352 0.0301 10.5078 10.0679 2.9459 0.5474 2.8796  ***************************************************************************** Limited Dependent Variable Model - CENSORED regression Maximum Likelihood Estimates Log-Likelihood.............. -700.72 Threshold values for the model: Lower 0.0000 Upper ********** N(0,1) used for significance levels. Variable Coefficient Std. Error t-ratio Prob. Var. Mean Constant 68.759 16.44 4.183 0.0000 J162 0.54184 0.3975 1.363 0.1729 0.3228 J177 0.41089 0.4848 0.848 0.3967 0.2336 J192 -0.40207 0.6611 -0.608 0.5431 0.2257 LDEPTH 11.118 2.323 4.786 0.0000 4.7937 L2DEP -0.99673 0.2202 -4.527 0.0000 23.3024 LGTEM 1.8515 0.9121 2.030 0.0424 1.7847 LSLOPE 0.30430 0.9130E-01 3.333 0.0009 3.2009 LMLD -1.9632 1.255 -1.564 0.1178 1.5147 LMSTEM 6.2444 3.290 1.898 0.0577 -0.0912 LMSAL -27.404 5.023 -5.456 0.0000 3.4544 LBMSAL -23.924 6.012 -3.979 0.0001 0.0403 HORVELM -0.11518 0.3585E-01 -3.213 0.0013 12.9840 MSHORVEL 0.88527E-01 0.3744E-01 2.364 0.0181 10.7954 BMHORVEL -0.10137 0.3833E-01 -2.645 0.0082 2.7609 LCHLA -0.28096 0.1695 -1.658 0.0973 0.7226 LMLD_DEP 0.34779 0.2720 1.278 0.2011 7.1489 Sigma 1.5348 0.5703E-01 26.914 0.0000  Var. st. dev. 0.4682 0.4237 0.4186 0.5688 5.3937 0.1747 0.9755 0.6330 0.0537 0.0352 0.0301 10.5078 10.0679 2.9459 0.5474 2.8796  Economic Valuation Of Critical Habitat Closures, Berman et al.  55  Flatfish, average weight > 0.5 kg Limited Dependent Variable Model - CENSORED Ordinary least squares regression. Observations 380 Mean of LHS 0.3833270E+01 StdDev of resid. 0.2825974E+01 R-squared 0.2266596E+00 F[ 18, 361] Log-likelihood -0.9242151E+03 Amemiya Pr. Criter. 0.8385436E+01 ANOVA Source Regression Residual Total N(0,1) used for significance levels. Variable Coefficient Std. Error Constant 2245.4 1293. J162 2.2061 0.9116 J177 2.9253 1.484 J192 4.1548 2.011 LDEPTH -10.339 4.653 L2DEP 1.3349 0.4847 LGTEM 71.379 17.96 LSLOPE 0.60310 0.1700 LMLD -1.5218 0.7217 LMTEM -4.4718 3.565 LMSTEM 19.905 6.696 LMSAL -1312.3 753.6 LMSSAL -25.502 14.33 LBMSAL -23.659 11.77 HORVELM -0.13246 0.6602E-01 MSHORVEL 0.11947 0.6950E-01 LCHLA1 0.70629 0.3459 L2GTEM -18.465 4.740 L2MSAL 188.68 109.7  regression Dep. Variable Weights Std.Dev of LHS Sum of squares Adj. R-squared 0.5878116E+01 Restr.(b=0) Log-l Akaike Info.Crit. Variation 0.8449811E+03 0.2882993E+04 0.3727974E+04 t-ratio 1.737 2.420 1.971 2.066 -2.222 2.754 3.974 3.548 -2.109 -1.254 2.973 -1.741 -1.779 -2.010 -2.006 1.719 2.042 -3.896 1.720  Prob. 0.0824 0.0155 0.0487 0.0389 0.0263 0.0059 0.0001 0.0004 0.0350 0.2097 0.0030 0.0816 0.0752 0.0444 0.0448 0.0856 0.0412 0.0001 0.0855  LBFLAT ONE 0.3136294E+01 0.2882993E+04 0.1880997E+00 Prob value -0.9730520E+03 0.4964290E+01 Deg. freedom 18. 361. 379.  0.1970590E-11 Mean Square 0.4694340E+02 0.7986130E+01 0.9836343E+01  Var. Mean  Var. st. dev.  0.3237 0.2342 0.2237 4.7950 23.3147 1.7844 3.1984 1.5198 1.9668 -0.0907 3.4550 0.0125 0.0399 13.0181 10.8239 0.9196 3.2144 11.9379  0.4685 0.4241 0.4173 0.5690 5.3955 0.1748 0.9756 0.6260 0.2118 0.0529 0.0336 0.0240 0.0287 10.5005 10.0659 0.4705 0.6562 0.2306  ***************************************************************************** Limited Dependent Variable Model - CENSORED regression Maximum Likelihood Estimates Log-Likelihood.............. -794.34 Threshold values for the model: Lower 0.0000 Upper ********** N(0,1) used for significance levels. Variable Coefficient Std. Error t-ratio Prob. Var. Mean Constant 3835.4 1951. 1.966 0.0493 J162 3.0538 1.387 2.202 0.0277 0.3237 J177 4.3090 2.286 1.885 0.0594 0.2342 J192 6.6923 3.094 2.163 0.0306 0.2237 LDEPTH -21.097 7.160 -2.947 0.0032 4.7950 L2DEP 2.5970 0.7463 3.480 0.0005 23.3147 LGTEM 125.55 28.99 4.330 0.0000 1.7844 LSLOPE 0.98501 0.2755 3.575 0.0003 3.1984 LMLD -2.5635 1.111 -2.308 0.0210 1.5198 LMTEM -7.6277 5.537 -1.378 0.1683 1.9668 LMSTEM 30.724 10.13 3.034 0.0024 -0.0907 LMSAL -2244.3 1138. -1.973 0.0485 3.4550 LMSSAL -42.533 21.92 -1.941 0.0523 0.0125 LBMSAL -41.613 18.34 -2.269 0.0233 0.0399 HORVELM -0.19004 0.1014 -1.874 0.0610 13.0181 MSHORVEL 0.17784 0.1062 1.674 0.0941 10.8239 LCHLA1 1.0526 0.5328 1.976 0.0482 0.9196 L2GTEM -32.519 7.659 -4.246 0.0000 3.2144 L2MSAL 323.27 165.6 1.952 0.0510 11.9379 Sigma 4.0471 0.2044 19.798 0.0000  Var. st. dev. 0.4685 0.4241 0.4173 0.5690 5.3955 0.1748 0.9756 0.6260 0.2118 0.0529 0.0336 0.0240 0.0287 10.5005 10.0659 0.4705 0.6562 0.2306  56  Economic Valuation Of Critical Habitat Closures, Berman et al.  Rockfish, all Limited Dependent Variable Model - CENSORED regression Ordinary least squares regression. Dep. Variable LROCK Observations 415 Weights ONE Mean of LHS 0.2230611E+01 Std.Dev of LHS 0.2518493E+01 StdDev of resid. 0.1838761E+01 Sum of squares 0.1359179E+04 R-squared 0.4823993E+00 Adj. R-squared 0.4669485E+00 F[ 12, 402] 0.3122170E+02 Prob value Log-likelihood -0.8350287E+03 Restr.(b=0) Log-l -0.9716781E+03 Amemiya Pr. Criter. 0.3486955E+01 Akaike Info.Crit. 0.4086885E+01 ANOVA Source Variation Deg. freedom Regression 0.1266743E+04 12. Residual 0.1359179E+04 402. Total 0.2625922E+04 414. N(0,1) used for significance levels. Variable Coefficient Std. Error t-ratio Prob. Var. Mean Constant 1422.6 369.8 3.847 0.0001 J162 -1.2637 0.3160 -3.999 0.0001 0.3108 J177 -1.3701 0.4226 -3.242 0.0012 0.2578 J192 -1.3041 0.5398 -2.416 0.0157 0.2217 LDEPTH -3.6561 2.092 -1.748 0.0805 4.7674 L2DEP 0.57280 0.2187 2.619 0.0088 23.0556 L2SLOPE -0.13411E-01 0.1962E-01 -0.684 0.4942 11.1722 LBMTEM -3.8807 1.039 -3.734 0.0002 -0.1073 LMSAL -846.88 216.5 -3.911 0.0001 3.4529 LBMSAL 10.041 5.904 1.701 0.0890 0.0399 VERTVEL -8.6183 4.686 -1.839 0.0659 -0.0024 BMHORVEL 0.52898E-01 0.4025E-01 1.314 0.1888 2.6973 L2MSAL 126.49 31.69 3.991 0.0001 11.9241 ***************************************************************************** Limited Dependent Variable Model - CENSORED regression Maximum Likelihood Estimates Log-Likelihood.............. -668.81 Threshold values for the model: Lower 0.0000 Upper ********** N(0,1) used for significance levels. Variable Coefficient Std. Error t-ratio Prob. Var. Mean Constant 1841.9 550.6 3.345 0.0008 J162 -1.8584 0.5091 -3.650 0.0003 0.3108 J177 -1.8095 0.7284 -2.484 0.0130 0.2578 J192 -2.2790 0.9834 -2.318 0.0205 0.2217 LDEPTH 11.207 4.418 2.537 0.0112 4.7674 L2DEP -0.84570 0.4490 -1.883 0.0596 23.0556 L2SLOPE -0.51155E-01 0.3195E-01 -1.601 0.1093 11.1722 LBMTEM -5.8966 1.814 -3.250 0.0012 -0.1073 LMSAL -1131.7 322.8 -3.506 0.0005 3.4529 LBMSAL 30.484 10.56 2.887 0.0039 0.0399 VERTVEL -12.762 6.933 -1.841 0.0656 -0.0024 BMHORVEL 0.14213 0.6724E-01 2.114 0.0345 2.6973 L2MSAL 170.48 47.33 3.602 0.0003 11.9241 Sigma 2.5557 0.1232 20.744 0.0000  0.2573607E-49 Mean Square 0.1055619E+03 0.3381043E+01 0.6342808E+01 Var. st. dev. 0.4634 0.4380 0.4159 0.5731 5.4133 5.3092 0.1828 0.0380 0.0320 0.0196 2.8813 0.2597  Var. st. dev. 0.4634 0.4380 0.4159 0.5731 5.4133 5.3092 0.1828 0.0380 0.0320 0.0196 2.8813 0.2597  Economic Valuation Of Critical Habitat Closures, Berman et al.  57  Rockfish, average weight > 0.5 kg Limited Dependent Variable Model - CENSORED Ordinary least squares regression. Observations 391 Mean of LHS 0.1339421E+01 StdDev of resid. 0.2223165E+01 R-squared 0.1678070E+00 F[ 15, 375] Log-likelihood -0.8590191E+03 Amemiya Pr. Criter. 0.5144713E+01 ANOVA Source Regression Residual Total N(0,1) used for significance levels. Variable Coefficient Std. Error Constant -64.247 20.25 J162 -0.74014 0.5727 J177 -0.74802 0.7823 J192 -1.8074 1.160 TIMELDEP 0.93623E-02 0.3178E-02 LSTEM 0.11830 1.023 LGTEM 11.193 10.23 L2SLOPE -0.22014E-01 0.2438E-01 LMLD 1.7334 1.323 LMSAL 14.083 6.198 LBMSAL 6.4959 7.952 VERTVEL -13.389 5.760 MSHORVEL -0.13167E-01 0.1447E-01 BMHORVEL 0.79452E-01 0.5379E-01 L2GTEM -3.0617 2.711 LMLD_DEP -0.45503 0.2926  regression Dep. Variable Weights Std.Dev of LHS Sum of squares Adj. R-squared 0.5041109E+01 Restr.(b=0) Log-l Akaike Info.Crit. Variation 0.3737325E+03 0.1853424E+04 0.2227156E+04 t-ratio -3.173 -1.292 -0.956 -1.558 2.946 0.116 1.094 -0.903 1.311 2.272 0.817 -2.324 -0.910 1.477 -1.130 -1.555  Prob. 0.0015 0.1962 0.3390 0.1193 0.0032 0.9079 0.2741 0.3664 0.1900 0.0231 0.4140 0.0201 0.3629 0.1397 0.2587 0.1199  LBROCK ONE 0.2389698E+01 0.1853424E+04 0.1345193E+00 Prob value -0.8949307E+03 0.4475801E+01 Deg. freedom 15. 375. 390.  0.3734648E-08 Mean Square 0.2491550E+02 0.4942464E+01 0.5710657E+01  Var. Mean  Var. st. dev.  0.3171 0.2302 0.2327 842.1405 2.2926 1.7858 11.2754 1.5073 3.4524 0.0410 -0.0024 10.5688 2.7515 3.2198 7.1040  0.4660 0.4215 0.4231 126.7421 0.1903 0.1758 5.3109 0.6411 0.0389 0.0323 0.0200 10.0444 2.9133 0.6594 2.9228  ***************************************************************************** Limited Dependent Variable Model - CENSORED regression Maximum Likelihood Estimates Log-Likelihood.............. -471.12 Threshold values for the model: Lower 0.0000 Upper ********** N(0,1) used for significance levels. Variable Coefficient Std. Error t-ratio Prob. Var. Mean Constant -410.09 86.33 -4.750 0.0000 J162 -1.9281 1.745 -1.105 0.2692 0.3171 J177 -1.3962 2.429 -0.575 0.5655 0.2302 J192 -7.0661 3.660 -1.931 0.0535 0.2327 TIMELDEP 0.37971E-01 0.1029E-01 3.689 0.0002 842.1405 LSTEM -6.2988 3.264 -1.930 0.0536 2.2926 LGTEM 171.57 49.20 3.487 0.0005 1.7858 L2SLOPE -0.87869E-01 0.7214E-01 -1.218 0.2232 11.2754 LMLD 8.5802 4.772 1.798 0.0722 1.5073 LMSAL 69.276 23.65 2.929 0.0034 3.4524 LBMSAL 58.733 29.96 1.961 0.0499 0.0410 VERTVEL -33.644 16.27 -2.068 0.0386 -0.0024 MSHORVEL -0.63866E-01 0.4067E-01 -1.570 0.1163 10.5688 BMHORVEL 0.30568 0.1795 1.703 0.0886 2.7515 L2GTEM -47.574 13.63 -3.491 0.0005 3.2198 LMLD_DEP -2.1410 1.009 -2.121 0.0339 7.1040 Sigma 5.0184 0.3764 13.334 0.0000  Var. st. dev. 0.4660 0.4215 0.4231 126.7421 0.1903 0.1758 5.3109 0.6411 0.0389 0.0323 0.0200 10.0444 2.9133 0.6594 2.9228  Economic Valuation Of Critical Habitat Closures, Berman et al.  58  APPENDIX B. EQUATIONS FOR SPATIAL DISTRIBUTION OF CATCH PER UNIT OF EFFORT (CPUE) FOR SUMMER AND WINTER, BERING SEA/ALEUTIAN ISLANDS AND GULF OF ALASKA, ESTIMATED FROM 2001 NMFS BOTTOM TRAWL FISHERIES OBSERVER DATA  1. Winter bottom trawl Selection equation, equations with chlorophyll (IMR2) Binomial Probit Model Maximum Likelihood Estimates Log-Likelihood.............. -803.90 Restricted (Slopes 0) Log-L. -4647.4 Chi-Squared (29)............ 7687.1 Significance Level.......... 0.32173E-13 N(0,1) used for significance levels. Variable Coefficient Std. Error Constant -8.0095 1.824 FEB -0.58440 0.2262 MAR -0.27818 0.2437 APR -0.33022 0.2787 NOV -0.51956 0.2642 DEC -1.3110 0.2641 GOA -0.28827 0.1897 LDEPTH 3.9029 0.6864 L2DEP -0.45071 0.7256E-01 SLOPE 0.25926E-01 0.1530E-01 SLOPE2 0.17093E-03 0.3905E-03 SST 0.21348 0.8084E-01 SST2 -0.50597E-01 0.1390E-01 SSTSLOPE 0.22275E-02 0.3028E-02 SSH -0.12662E-01 0.1032E-01 SSHSLOPE -0.20823E-02 0.2055E-02 MWIND -0.43418 0.3434 MCHLA -0.30384 0.1549 MCHLA1 0.16545 0.1648 DWIND -0.32717 0.1485 DCHLA -0.38978 0.1410 DCHLA1 -0.27557 0.1247 POLTRAWL -0.18045 0.1398 CODTRAWL 0.64341 0.1811 ATKTRAWL 1.4558 0.3287 POLTSSL 0.23449E-01 0.1772 CODTSSL 0.48325 0.2669 ATKTSSL -0.84069 0.3077 MIXTSSL -1.1763 0.3151 PORTDIST -0.17414E-01 0.2034E-02 Frequencies of actual & predicted outcomes Predicted outcome has maximum probability.  t-ratio -4.392 -2.583 -1.141 -1.185 -1.967 -4.963 -1.520 5.686 -6.212 1.694 0.438 2.641 -3.641 0.736 -1.227 -1.013 -1.264 -1.962 1.004 -2.204 -2.765 -2.211 -1.291 3.552 4.429 0.132 1.810 -2.732 -3.733 -8.562  Predicted Actual  0  1  TOTAL  0 1  9161 1652  3 13  9164 1665  Total  10813  16  10829  Prob. 0.0000 0.0098 0.2537 0.2361 0.0492 0.0000 0.1285 0.0000 0.0000 0.0902 0.6616 0.0083 0.0003 0.4620 0.2198 0.3110 0.2061 0.0498 0.3153 0.0276 0.0057 0.0271 0.1967 0.0004 0.0000 0.8947 0.0702 0.0063 0.0002 0.0000  Var. Mean  Var. st. dev.  0.1770 0.1625 0.1592 0.1691 0.1633 0.2505 4.4569 20.7804 3.9482 84.4372 2.4452 11.6614 13.8289 -3.4514 32.0651 2.4399 0.3811 0.4223 0.2554 0.4946 0.6494 0.4259 0.7303 0.0193 0.1648 0.1515 0.1387 0.0913 63.9429  0.3817 0.3689 0.3659 0.3748 0.3697 0.4333 0.9573 8.8173 8.2979 287.8697 2.6567 14.7372 13.0156 7.7605 19.2719 0.1314 0.3658 0.3927 0.4361 0.5000 0.4772 0.4510 0.3938 0.1173 0.3710 0.3577 0.3448 0.2871 39.7322  Economic Valuation Of Critical Habitat Closures, Berman et al.  59  Selection equation, equations without chlorophyll (IMR3) Binomial Probit Model Maximum Likelihood Estimates Log-Likelihood.............. -827.24 Restricted (Slopes 0) Log-L. -4125.9 Chi-Squared (24)............ 6597.3 Significance Level.......... 0.32173E-13 N(0,1) used for significance levels. Variable Coefficient Std. Error Constant -8.2873 1.709 FEB -0.36374 0.1918 MAR 0.20799 0.1758 APR 0.19676 0.1960 NOV -0.46857E-01 0.1865 DEC -0.73421 0.2215 GOA -0.16994 0.1915 LDEPTH 3.6863 0.6554 L2DEP -0.42507 0.6937E-01 SLOPE 0.23553E-01 0.1582E-01 SLOPE2 0.22750E-03 0.4127E-03 SST 0.29195 0.8387E-01 SST2 -0.62001E-01 0.1418E-01 SSTSLOPE 0.29445E-02 0.3113E-02 SSH -0.13987E-01 0.1072E-01 SSHSLOPE -0.29155E-02 0.2061E-02 LWIND -0.45863 0.3288 POLTRAWL -0.13340 0.1389 CODTRAWL 0.58830 0.1838 ATKTRAWL 1.7320 0.3335 POLTSSL 0.58119E-01 0.1852 CODTSSL 0.35259 0.2944 ATKTSSL -0.69082 0.3294 MIXTSSL -1.1321 0.3149 PORTDIST -0.18465E-01 0.2015E-02 Frequencies of actual & predicted outcomes Predicted outcome has maximum probability.  t-ratio -4.848 -1.897 1.183 1.004 -0.251 -3.314 -0.887 5.624 -6.127 1.489 0.551 3.481 -4.371 0.946 -1.304 -1.415 -1.395 -0.960 3.201 5.194 0.314 1.198 -2.097 -3.595 -9.165  Predicted Actual  0  1  TOTAL  0 1  7101 1577  3 7  7104 1584  Total  8678  10  8688  Prob. 0.0000 0.0579 0.2369 0.3154 0.8016 0.0009 0.3749 0.0000 0.0000 0.1366 0.5815 0.0005 0.0000 0.3442 0.1922 0.1572 0.1631 0.3368 0.0014 0.0000 0.7536 0.2311 0.0360 0.0003 0.0000  Var. Mean  Var. st. dev.  0.1912 0.1597 0.1459 0.1877 0.1502 0.2269 4.6349 22.2899 4.6016 103.5289 2.9045 12.8857 13.5910 -3.2951 30.6658 2.4428 0.4196 0.7504 0.0221 0.1396 0.1300 0.1257 0.0785 59.9196  0.3933 0.3663 0.3530 0.3905 0.3573 0.4189 0.8986 8.6848 9.0754 316.6736 2.1915 13.7415 12.1874 6.6821 18.2006 0.1455 0.4519 0.3823 0.1248 0.3466 0.3351 0.3305 0.2675 38.0269  60  Economic Valuation Of Critical Habitat Closures, Berman et al.  Pollock, standard CPUE Sample Selection Model Two stage least squares regression. Dep. Variable LOGPOLLS Observations 1653 Weights ONE Mean of LHS 0.4938884E+00 Std.Dev of LHS StdDev of resid. 0.5295141E+00 Sum of squares 0.4547848E+03 R-squared 0.2022732E+00 Adj. R-squared 0.1875187E+00 F[ 30, 1622] 0.1370925E+02 Prob value 0.3217295E-13 Log-likelihood -0.1278888E+04 Restr.(b=0) Log-l Amemiya Pr. Criter. 0.2856435E+00 Akaike Info.Crit. Standard error corrected for selection..... 0.52995 Correlation of disturbance in regression and Selection Criterion (Rho).............. 0.44612E-01 N(0,1) used for significance levels. Variable Coefficient Std. Error t-ratio Prob. Constant 1.1154 1.140 0.979 0.3278 FEB -0.15042 0.1829 -0.822 0.4108 MAR -0.37005 0.3468 -1.067 0.2859 APR -0.65578 0.5082 -1.290 0.1969 NOV -2.2821 1.489 -1.532 0.1255 DEC -2.4468 1.622 -1.509 0.1314 GOA -1.3082 1.051 -1.245 0.2130 TIMELDEP 0.50111E-01 0.3452E-01 1.452 0.1466 LDEPTH -0.36414 0.4314 -0.844 0.3986 L2DEP -0.46261E-03 0.4333E-01 -0.011 0.9915 SLOPE -0.25176E-01 0.5478E-02 -4.595 0.0000 SLOPE2 0.37079E-03 0.1313E-03 2.825 0.0047 SST 0.16426 0.4515E-01 3.638 0.0003 SST2 -0.30844E-01 0.1080E-01 -2.856 0.0043 SSTSLOPE -0.28840E-02 0.1265E-02 -2.279 0.0227 SSH 0.12895E-01 0.5204E-02 2.478 0.0132 SSHSLOPE -0.95708E-03 0.8205E-03 -1.166 0.2434 MWIND 0.34503 0.1548 2.229 0.0258 MCHLA 0.87458E-01 0.9232E-01 0.947 0.3435 MCHLA1 -0.80405E-01 0.9173E-01 -0.877 0.3807 DWIND 0.12052 0.6758E-01 1.783 0.0745 DCHLA -0.12035 0.6483E-01 -1.856 0.0634 DCHLA1 -0.30161E-02 0.5271E-01 -0.057 0.9544 GLDEPTH 0.93711 0.1131 8.286 0.0000 GTIMELDE -0.86798E-01 0.2006E-01 -4.326 0.0000 GSST -0.85188E-01 0.9814E-01 -0.868 0.3854 GSSH -0.27424E-01 0.1321E-01 -2.075 0.0380 GMWIND -0.79279 0.3654 -2.170 0.0300 GMCHLA -0.12001 0.1420 -0.845 0.3981 GMCHLA1 0.34060 0.3861 0.882 0.3777 IMR2 0.23642E-01 0.5027E-01 0.470 0.6381  0.5930371E+00  -0.1481315E+04 0.1584862E+01  Var. Mean  Var. st. dev.  0.1506 0.3315 0.2989 0.0901 0.0321 0.1688 17.6167 4.5252 20.7261 3.9975 95.6138 3.1065 11.2998 13.1123 -5.3569 26.4713 2.3861 0.4752 0.4445 0.0490 0.2390 0.3696 0.8006 2.3168 0.8209 -0.6570 0.3997 0.0888 0.0497 1.8333  0.3578 0.4709 0.4579 0.2865 0.1762 0.3747 11.9465 0.4984 4.8625 8.9265 342.6782 1.3323 7.7299 11.4851 5.1763 17.4774 0.1216 0.3091 0.3390 0.2159 0.4266 0.4829 1.7913 5.6658 1.8344 2.5355 0.8888 0.2504 0.1298 0.4828  Economic Valuation Of Critical Habitat Closures, Berman et al.  61  Pollock, standard CPUE (cont.) Limited Dependent Variable Model - CENSORED Maximum Likelihood Estimates Log-Likelihood.............. -1398.2 Threshold values for the model: Lower N(0,1) used for significance levels. Variable Coefficient Std. Error Constant 1.0848 1.336 FEB -0.25756 0.2090 MAR -0.65751 0.3979 APR -1.0791 0.5832 NOV -3.7238 1.715 DEC -4.0426 1.869 GOA -2.3676 1.303 TIMELDEP 0.85701E-01 0.3983E-01 LDEPTH -0.38976 0.5074 L2DEP -0.69750E-02 0.5060E-01 SLOPE -0.41843E-01 0.6649E-02 SLOPE2 0.56804E-03 0.1606E-03 SST 0.16687 0.5200E-01 SST2 -0.34356E-01 0.1227E-01 SSTSLOPE -0.33352E-02 0.1474E-02 SSH 0.16015E-01 0.6061E-02 SSHSLOPE -0.10924E-02 0.9511E-03 MWIND 0.43538 0.1826 MCHLA 0.98047E-01 0.1037 MCHLA1 -0.74557E-01 0.1026 DWIND 0.10749 0.8102E-01 DCHLA -0.14061 0.7484E-01 DCHLA1 0.14144E-01 0.5972E-01 GLDEPTH 1.5109 0.1392 GTIMELDE -0.14220 0.2449E-01 GSST -0.27316 0.1174 GSSH -0.47740E-01 0.1682E-01 GMWIND -0.90118 0.4478 GMCHLA -0.27502 0.2025 GMCHLA1 0.58058 0.4593 IMR2 -0.31325E-01 0.5813E-01 Sigma 0.58721 0.1127E-01  regression 0.0000  Upper  **********  t-ratio 0.812 -1.232 -1.652 -1.850 -2.171 -2.162 -1.817 2.152 -0.768 -0.138 -6.293 3.537 3.209 -2.800 -2.262 2.642 -1.149 2.385 0.945 -0.727 1.327 -1.879 0.237 10.853 -5.807 -2.326 -2.838 -2.013 -1.358 1.264 -0.539 52.082  Prob. 0.4167 0.2179 0.0985 0.0643 0.0299 0.0306 0.0692 0.0314 0.4424 0.8904 0.0000 0.0004 0.0013 0.0051 0.0237 0.0082 0.2507 0.0171 0.3445 0.4672 0.1846 0.0603 0.8128 0.0000 0.0000 0.0200 0.0045 0.0442 0.1744 0.2062 0.5900 0.0000  Var. Mean  Var. st. dev.  0.1508 0.3317 0.2979 0.0899 0.0320 0.1683 17.5926 4.5262 20.7345 4.0358 96.9843 3.1077 11.3046 13.0894 -5.3576 26.5137 2.3863 0.4746 0.4437 0.0489 0.2394 0.3703 0.7982 2.3099 0.8184 -0.6551 0.3985 0.0885 0.0496 1.8289  0.3579 0.4710 0.4575 0.2861 0.1760 0.3742 11.9387 0.4981 4.8581 8.9858 345.8335 1.3310 7.7231 11.4765 5.1789 17.5059 0.1215 0.3089 0.3389 0.2156 0.4269 0.4830 1.7891 5.6586 1.8322 2.5319 0.8878 0.2501 0.1296 0.4856  Economic Valuation Of Critical Habitat Closures, Berman et al.  62  Pollock, standard CPUE (cont.) ML Estimates of Selection Model Maximum Likelihood Estimates Log-Likelihood.............. -4626.4 LHS is CENSORED. Tobit Model fit by MLE. FIRST 30 estimates are probit equation. N(0,1) used for significance levels. Variable Coefficient Std. Error Constant -4.1264 1.232 FEB 0.98613E-01 0.1709 MAR 0.37777 0.3067 APR 0.58488 0.4458 NOV 1.1308 1.270 DEC 1.3228 1.384 GOA -0.14458 1.335 TIMELDEP -0.34752E-01 0.2935E-01 LDEPTH 1.7974 0.4519 L2DEP -0.22567 0.4269E-01 SLOPE -0.19990E-01 0.8378E-02 SLOPE2 0.56498E-03 0.2294E-03 SST 0.36798 0.4871E-01 SST2 -0.53505E-01 0.1187E-01 SSTSLOPE 0.14164E-03 0.1674E-02 SSH 0.21621E-01 0.6149E-02 SSHSLOPE -0.33315E-02 0.1036E-02 MWIND 0.53113E-01 0.2104 MCHLA 0.12475E-01 0.1135 MCHLA1 0.19246 0.1211 DWIND -0.18806 0.7618E-01 DCHLA -0.37030 0.7953E-01 DCHLA1 -0.27602 0.6146E-01 GLDEPTH 1.1358 0.1040 GTIMELDE -0.11683 0.2030E-01 GSST -0.35817 0.8837E-01 GSSH -0.43281E-01 0.1345E-01 GMWIND -1.0219 0.4806 GMCHLA -0.11160 0.2064 GMCHLA1 -0.41817 0.3863 SIGMA(1) 0.88965 0.1663E-01 RHO(1,2) 0.97103 0.5738E-02  t-ratio -3.349 0.577 1.232 1.312 0.891 0.956 -0.108 -1.184 3.977 -5.287 -2.386 2.463 7.554 -4.509 0.085 3.516 -3.214 0.252 0.110 1.590 -2.468 -4.656 -4.491 10.921 -5.755 -4.053 -3.217 -2.126 -0.541 -1.082 53.499 169.218  Prob. 0.0008 0.5639 0.2181 0.1895 0.3731 0.3393 0.9138 0.2364 0.0001 0.0000 0.0170 0.0138 0.0000 0.0000 0.9325 0.0004 0.0013 0.8007 0.9125 0.1119 0.0136 0.0000 0.0000 0.0000 0.0000 0.0001 0.0013 0.0335 0.5888 0.2791 0.0000 0.0000  Var. Mean  Var. st. dev.  Economic Valuation Of Critical Habitat Closures, Berman et al.  63  Pacific Cod, CPUE Sample Selection Model Two stage least squares regression. Dep. Variable LOGPCOD Observations 1653 Weights ONE Mean of LHS 0.4016436E+00 Std.Dev of LHS 0.4578468E+00 StdDev of resid. 0.3778806E+00 Sum of squares R-squared 0.3183967E+00 Adj. R-squared 0.3057900E+00 F[ 30, 1622] 0.2525611E+02 Prob value 0.3217295E-13 Log-likelihood -0.7211968E+03 Restr.(b=0) Log-l Amemiya Pr. Criter. 0.1454717E+00 Akaike Info.Crit. Standard error corrected for selection..... 0.37804 Correlation of disturbance in regression and Selection Criterion (Rho).............. -0.34165E-01 N(0,1) used for significance levels. Variable Coefficient Std. Error t-ratio Prob. Constant -0.34035 0.8704 -0.391 0.6958 FEB 0.26840 0.1298 2.067 0.0387 MAR 0.27857 0.2469 1.128 0.2592 APR 0.21542 0.3635 0.593 0.5535 NOV -0.38747 1.065 -0.364 0.7159 DEC -0.59297 1.157 -0.512 0.6084 GOA 0.60395 0.7511 0.804 0.4214 TIMELDEP 0.87670E-02 0.2468E-01 0.355 0.7225 LDEPTH 0.60747 0.3311 1.835 0.0665 L2DEP -0.91019E-01 0.3371E-01 -2.700 0.0069 SLOPE 0.31871E-01 0.4234E-02 7.528 0.0000 SLOPE2 -0.56813E-03 0.9254E-04 -6.140 0.0000 SST -0.26935 0.3413E-01 -7.892 0.0000 SST2 0.43395E-01 0.7856E-02 5.524 0.0000 SSTSLOPE -0.15933E-02 0.9065E-03 -1.758 0.0788 SSH 0.16291E-01 0.3711E-02 4.390 0.0000 SSHSLOPE 0.14028E-02 0.5978E-03 2.347 0.0189 MWIND 0.11299 0.1097 1.030 0.3032 MCHLA -0.43405 0.6560E-01 -6.616 0.0000 MCHLA1 -0.19937 0.6630E-01 -3.007 0.0026 DWIND 0.16400 0.4901E-01 3.347 0.0008 DCHLA 0.11784E-01 0.4757E-01 0.248 0.8043 DCHLA1 -0.79232E-01 0.3865E-01 -2.050 0.0404 GLDEPTH -0.45225 0.8058E-01 -5.612 0.0000 GTIMELDE 0.16988E-01 0.1431E-01 1.187 0.2351 GSST 0.61049E-01 0.6931E-01 0.881 0.3784 GSSH 0.28799E-02 0.9423E-02 0.306 0.7599 GMWIND 0.39574 0.2607 1.518 0.1291 GMCHLA 0.24716 0.1015 2.436 0.0149 GMCHLA1 -0.45681 0.2783 -1.642 0.1007 IMR2 -0.12916E-01 0.6029E-01 -0.214 0.8304  0.2316115E+03 -0.1053648E+04 0.9100990E+00  Var. Mean  Var. st. dev.  0.1506 0.3315 0.2989 0.0901 0.0321 0.1688 17.6167 4.5252 20.7261 3.9975 95.6138 3.1065 11.2998 13.1123 -5.3569 26.4713 2.3861 0.4752 0.4445 0.0490 0.2390 0.3696 0.8006 2.3168 0.8209 -0.6570 0.3997 0.0888 0.0497 1.1629  0.3578 0.4709 0.4579 0.2865 0.1762 0.3747 11.9465 0.4984 4.8625 8.9265 342.6782 1.3323 7.7299 11.4851 5.1763 17.4774 0.1216 0.3091 0.3390 0.2159 0.4266 0.4829 1.7913 5.6658 1.8344 2.5355 0.8888 0.2504 0.1298 0.3469  Economic Valuation Of Critical Habitat Closures, Berman et al.  64  Pacific Cod, CPUE (cont.) Limited Dependent Variable Model - CENSORED Maximum Likelihood Estimates Log-Likelihood.............. -836.10 Threshold values for the model: Lower N(0,1) used for significance levels. Variable Coefficient Std. Error Constant -4.5811 0.9639 FEB 0.45632 0.1401 MAR 0.63246 0.2668 APR 0.75070 0.3913 NOV 1.1011 1.152 DEC 1.0698 1.255 GOA 1.5165 0.8081 TIMELDEP -0.26526E-01 0.2670E-01 LDEPTH 2.3686 0.3710 L2DEP -0.27027 0.3711E-01 SLOPE 0.30314E-01 0.4177E-02 SLOPE2 -0.50497E-03 0.9921E-04 SST -0.24102 0.3388E-01 SST2 0.39899E-01 0.8133E-02 SSTSLOPE -0.94942E-03 0.9545E-03 SSH 0.20636E-01 0.3927E-02 SSHSLOPE 0.10670E-02 0.6211E-03 MWIND 0.80353E-01 0.1154 MCHLA -0.45601 0.6975E-01 MCHLA1 -0.18075 0.6914E-01 DWIND 0.15257 0.5143E-01 DCHLA -0.28402E-01 0.4859E-01 DCHLA1 -0.12141 0.3971E-01 GLDEPTH -0.71826 0.9939E-01 GTIMELDE 0.44300E-02 0.1623E-01 GSST 0.73247E-01 0.7624E-01 GSSH -0.55089E-02 0.1040E-01 GMWIND 0.49946 0.2768 GMCHLA 0.34294 0.1083 GMCHLA1 -0.46969 0.3011 IMR2 0.93349E-01 0.3800E-01 Sigma 0.39417 0.7186E-02  regression 0.0000  Upper  **********  t-ratio -4.753 3.257 2.370 1.919 0.956 0.853 1.877 -0.994 6.385 -7.283 7.257 -5.090 -7.114 4.906 -0.995 5.255 1.718 0.696 -6.537 -2.614 2.966 -0.585 -3.057 -7.226 0.273 0.961 -0.530 1.805 3.166 -1.560 2.457 54.851  Prob. 0.0000 0.0011 0.0178 0.0550 0.3393 0.3938 0.0606 0.3204 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.3199 0.0000 0.0858 0.4863 0.0000 0.0089 0.0030 0.5588 0.0022 0.0000 0.7849 0.3367 0.5963 0.0711 0.0015 0.1188 0.0140 0.0000  Var. Mean  Var. st. dev.  0.1510 0.3309 0.2983 0.0900 0.0320 0.1685 17.5967 4.5260 20.7325 4.0364 97.0856 3.1073 11.3035 13.0978 -5.3587 26.4999 2.3863 0.4747 0.4438 0.0489 0.2397 0.3708 0.7991 2.3127 0.8194 -0.6558 0.3990 0.0886 0.0496 1.8293  0.3581 0.4707 0.4577 0.2862 0.1761 0.3744 11.9453 0.4983 4.8607 8.9912 346.0301 1.3317 7.7274 11.4800 5.1819 17.4975 0.1216 0.3091 0.3390 0.2158 0.4271 0.4832 1.7900 5.6615 1.8331 2.5334 0.8882 0.2502 0.1297 0.4856  Economic Valuation Of Critical Habitat Closures, Berman et al.  65  Pacific Cod, CPUE (cont.) ML Estimates of Selection Model Maximum Likelihood Estimates Log-Likelihood.............. -4088.5 LHS is CENSORED. Tobit Model fit by MLE. FIRST 30 estimates are probit equation. N(0,1) used for significance levels. Variable Coefficient Std. Error Constant -7.2310 0.8258 FEB 0.45676 0.1256 MAR 0.82684 0.2313 APR 1.1856 0.3407 NOV 2.5329 0.9725 DEC 2.6828 1.051 GOA 0.73664 0.6481 TIMELDEP -0.64675E-01 0.2215E-01 LDEPTH 3.4550 0.3167 L2DEP -0.38971 0.2937E-01 SLOPE 0.37766E-01 0.4092E-02 SLOPE2 -0.44846E-03 0.9988E-04 SST 0.77759E-02 0.2956E-01 SST2 -0.10414E-04 0.8001E-02 SSTSLOPE 0.55426E-03 0.9146E-03 SSH 0.21176E-01 0.4207E-02 SSHSLOPE -0.41704E-03 0.6674E-03 MWIND -0.69156E-01 0.1025 MCHLA -0.38958 0.7398E-01 MCHLA1 -0.38286E-02 0.9325E-01 DWIND -0.71155E-01 0.4193E-01 DCHLA -0.16925 0.4913E-01 DCHLA1 -0.25188 0.4418E-01 GLDEPTH -0.37785 0.7760E-01 GTIMELDE 0.16939E-01 0.1290E-01 GSST -0.15007E-01 0.5949E-01 GSSH 0.13846E-01 0.7467E-02 GMWIND 0.40362 0.2273 GMCHLA 0.14655 0.8716E-01 GMCHLA1 -0.79825 0.2400 SIGMA(1) 0.60701 0.1138E-01 RHO(1,2) 0.96156 0.5675E-02  t-ratio -8.756 3.637 3.575 3.480 2.605 2.553 1.137 -2.920 10.908 -13.270 9.228 -4.490 0.263 -0.001 0.606 5.033 -0.625 -0.675 -5.266 -0.041 -1.697 -3.445 -5.702 -4.869 1.313 -0.252 1.854 1.776 1.681 -3.326 53.345 169.443  Prob. 0.0000 0.0003 0.0004 0.0005 0.0092 0.0107 0.2557 0.0035 0.0000 0.0000 0.0000 0.0000 0.7925 0.9990 0.5445 0.0000 0.5321 0.5000 0.0000 0.9673 0.0897 0.0006 0.0000 0.0000 0.1892 0.8008 0.0637 0.0757 0.0927 0.0009 0.0000 0.0000  Var. Mean  Var. st. dev.  66  Economic Valuation Of Critical Habitat Closures, Berman et al.  Pacific cod, standard CPUE Sample Selection Model Two stage least squares regression. Dep. Variable LOGPCODS Observations 1653 Weights ONE Mean of LHS 0.5852257E+00 Std.Dev of LHS 0.6311426E+00 StdDev of resid. 0.4941459E+00 Sum of squares 0.3960602E+03 R-squared 0.3866361E+00 Adj. R-squared 0.3752915E+00 F[ 30, 1622] 0.3408112E+02 Prob value 0.3217295E-13 Log-likelihood -0.1164618E+04 Restr.(b=0) Log-l Amemiya Pr. Criter. 0.2487594E+00 Akaike Info.Crit. Standard error corrected for selection..... 0.51396 Correlation of disturbance in regression and Selection Criterion (Rho).............. 0.32420 N(0,1) used for significance levels. Variable Coefficient Std. Error t-ratio Prob. Constant -1.8615 1.131 -1.646 0.0997 FEB 0.23511 0.1699 1.384 0.1665 MAR -0.50209E-01 0.3220 -0.156 0.8761 APR -0.34159 0.4737 -0.721 0.4709 NOV -2.3463 1.384 -1.695 0.0901 DEC -2.6294 1.505 -1.747 0.0806 GOA 2.2027 0.9816 2.244 0.0248 TIMELDEP 0.47112E-01 0.3206E-01 1.470 0.1417 LDEPTH 1.3597 0.4333 3.138 0.0017 L2DEP -0.19238 0.4435E-01 -4.338 0.0000 SLOPE 0.32883E-01 0.5784E-02 5.686 0.0000 SLOPE2 -0.54462E-03 0.1239E-03 -4.396 0.0000 SST -0.25440 0.4350E-01 -5.848 0.0000 SST2 0.41775E-01 0.1022E-01 4.087 0.0000 SSTSLOPE -0.24015E-02 0.1207E-02 -1.989 0.0467 SSH 0.32392E-01 0.4910E-02 6.597 0.0000 SSHSLOPE 0.21011E-02 0.7808E-03 2.691 0.0071 MWIND 0.16945 0.1458 1.163 0.2450 MCHLA -0.48255 0.8668E-01 -5.567 0.0000 MCHLA1 -0.17339 0.8625E-01 -2.010 0.0444 DWIND 0.91711E-01 0.6415E-01 1.430 0.1528 DCHLA -0.98280E-01 0.6258E-01 -1.570 0.1163 DCHLA1 -0.18044 0.5224E-01 -3.454 0.0006 GLDEPTH -0.59690 0.1053 -5.668 0.0000 GTIMELDE 0.15717E-01 0.1872E-01 0.840 0.4011 GSST -0.79708E-01 0.9060E-01 -0.880 0.3790 GSSH 0.11185E-01 0.1230E-01 0.909 0.3633 GMWIND 0.41358 0.3414 1.211 0.2258 GMCHLA 0.62346E-01 0.1325 0.471 0.6380 GMCHLA1 -0.32412 0.3622 -0.895 0.3709 IMR2 0.16663 0.7790E-01 2.139 0.0324  -0.1584256E+04 0.1446604E+01  Var. Mean  Var. st. dev.  0.1506 0.3315 0.2989 0.0901 0.0321 0.1688 17.6167 4.5252 20.7261 3.9975 95.6138 3.1065 11.2998 13.1123 -5.3569 26.4713 2.3861 0.4752 0.4445 0.0490 0.2390 0.3696 0.8006 2.3168 0.8209 -0.6570 0.3997 0.0888 0.0497 1.1639  0.3578 0.4709 0.4579 0.2865 0.1762 0.3747 11.9465 0.4984 4.8625 8.9265 342.6782 1.3323 7.7299 11.4851 5.1763 17.4774 0.1216 0.3091 0.3390 0.2159 0.4266 0.4829 1.7913 5.6658 1.8344 2.5355 0.8888 0.2504 0.1298 0.3593  Economic Valuation Of Critical Habitat Closures, Berman et al.  67  Pacific cod, standard CPUE (cont.) Limited Dependent Variable Model - CENSORED Maximum Likelihood Estimates Log-Likelihood.............. -1251.5 Threshold values for the model: Lower N(0,1) used for significance levels. Variable Coefficient Std. Error Constant -6.3956 1.267 FEB 0.45232 0.1837 MAR 0.34229 0.3499 APR 0.24404 0.5131 NOV -0.61666 1.511 DEC -0.69585 1.646 GOA 3.2275 1.058 TIMELDEP 0.79667E-02 0.3502E-01 LDEPTH 3.2265 0.4874 L2DEP -0.37358 0.4863E-01 SLOPE 0.24030E-01 0.5474E-02 SLOPE2 -0.43733E-03 0.1301E-03 SST -0.25286 0.4443E-01 SST2 0.40570E-01 0.1066E-01 SSTSLOPE -0.20002E-02 0.1252E-02 SSH 0.36651E-01 0.5154E-02 SSHSLOPE 0.20999E-02 0.8134E-03 MWIND 0.18850 0.1513 MCHLA -0.47418 0.9140E-01 MCHLA1 -0.19012 0.9065E-01 DWIND 0.11420 0.6719E-01 DCHLA -0.97964E-01 0.6375E-01 DCHLA1 -0.16553 0.5207E-01 GLDEPTH -0.92506 0.1280 GTIMELDE -0.69064E-02 0.2118E-01 GSST -0.63529E-01 0.9966E-01 GSSH -0.16994E-02 0.1360E-01 GMWIND 0.60086 0.3624 GMCHLA 0.19110 0.1419 GMCHLA1 -0.67475E-01 0.3927 IMR2 0.81921E-01 0.4980E-01 Sigma 0.51663 0.9425E-02  regression 0.0000  Upper  **********  t-ratio -5.048 2.462 0.978 0.476 -0.408 -0.423 3.050 0.228 6.620 -7.681 4.390 -3.363 -5.691 3.804 -1.597 7.111 2.582 1.246 -5.188 -2.097 1.700 -1.537 -3.179 -7.229 -0.326 -0.637 -0.125 1.658 1.347 -0.172 1.645 54.815  Prob. 0.0000 0.0138 0.3280 0.6344 0.6833 0.6724 0.0023 0.8200 0.0000 0.0000 0.0000 0.0008 0.0000 0.0001 0.1102 0.0000 0.0098 0.2127 0.0000 0.0360 0.0892 0.1244 0.0015 0.0000 0.7443 0.5238 0.9005 0.0973 0.1779 0.8636 0.1000 0.0000  Var. Mean  Var. st. dev.  0.1510 0.3309 0.2983 0.0900 0.0320 0.1685 17.5967 4.5260 20.7325 4.0364 97.0856 3.1073 11.3035 13.0978 -5.3587 26.4999 2.3863 0.4747 0.4438 0.0489 0.2397 0.3708 0.7991 2.3127 0.8194 -0.6558 0.3990 0.0886 0.0496 1.8293  0.3581 0.4707 0.4577 0.2862 0.1761 0.3744 11.9453 0.4983 4.8607 8.9912 346.0301 1.3317 7.7274 11.4800 5.1819 17.4975 0.1216 0.3091 0.3390 0.2158 0.4271 0.4832 1.7900 5.6615 1.8331 2.5334 0.8882 0.2502 0.1297 0.4856  Economic Valuation Of Critical Habitat Closures, Berman et al.  68  Pacific cod, standard CPUE (cont.) ML Estimates of Selection Model Maximum Likelihood Estimates Log-Likelihood.............. -4566.3 LHS is CENSORED. Tobit Model fit by MLE. FIRST 30 estimates are probit equation. N(0,1) used for significance levels. Variable Coefficient Std. Error Constant -6.3867 1.286 FEB 0.45635 0.2193 MAR 0.37182 0.4336 APR 0.31363 0.6440 NOV -0.32962 1.885 DEC -0.38884 2.036 GOA 3.3144 0.9396 TIMELDEP 0.14914E-02 0.4311E-01 LDEPTH 3.2197 0.4993 L2DEP -0.36956 0.4759E-01 SLOPE 0.25667E-01 0.4666E-02 SLOPE2 -0.47429E-03 0.1005E-03 SST -0.26058 0.4351E-01 SST2 0.41794E-01 0.1195E-01 SSTSLOPE -0.20963E-02 0.1247E-02 SSH 0.35638E-01 0.5527E-02 SSHSLOPE 0.21735E-02 0.8050E-03 MWIND 0.20655 0.1310 MCHLA -0.45320 0.1115 MCHLA1 -0.19234 0.1506 DWIND 0.11284 0.4940E-01 DCHLA -0.90226E-01 0.6071E-01 DCHLA1 -0.15534 0.5928E-01 GLDEPTH -0.94200 0.1275 GTIMELDE -0.40522E-02 0.2032E-01 GSST -0.66750E-01 0.9346E-01 GSSH -0.28497E-03 0.1100E-01 GMWIND 0.60259 0.3084 GMCHLA 0.19075 0.1428 GMCHLA1 -0.87730E-01 0.3621 SIGMA(1) 0.51882 0.8805E-02 RHO(1,2) 0.13195 0.1042  t-ratio -4.966 2.081 0.858 0.487 -0.175 -0.191 3.527 0.035 6.448 -7.765 5.501 -4.718 -5.989 3.497 -1.681 6.448 2.700 1.576 -4.064 -1.277 2.284 -1.486 -2.621 -7.389 -0.199 -0.714 -0.026 1.954 1.336 -0.242 58.922 1.266  Prob. 0.0000 0.0375 0.3911 0.6263 0.8612 0.8486 0.0004 0.9724 0.0000 0.0000 0.0000 0.0000 0.0000 0.0005 0.0927 0.0000 0.0069 0.1149 0.0000 0.2014 0.0224 0.1372 0.0088 0.0000 0.8420 0.4751 0.9793 0.0507 0.1817 0.8085 0.0000 0.2054  Var. Mean  Var. st. dev.  Economic Valuation Of Critical Habitat Closures, Berman et al.  69  Atka mackerel, standard CPUE Sample Selection Model Two stage least squares regression. Dep. Variable LOGATKAS Observations 1581 Weights ONE Mean of LHS 0.9712528E-01 Std.Dev of LHS 0.5179737E+00 StdDev of resid. 0.3773482E+00 Sum of squares 0.2217039E+03 R-squared 0.4770010E+00 Adj. R-squared 0.4692753E+00 F[ 23, 1557] 0.6174180E+02 Prob value 0.3217295E-13 Log-likelihood -0.6904281E+03 Restr.(b=0) Log-l Amemiya Pr. Criter. 0.1445532E+00 Akaike Info.Crit. ANOVA Source Variation Deg. freedom Regression 0.2022049E+03 23. Residual 0.2217039E+03 1557. Total 0.4239088E+03 1580. N(0,1) used for significance levels. Variable Coefficient Std. Error t-ratio Prob. Constant -2.5075 0.7676 -3.267 0.0011 FEB 0.36633 0.1212 3.023 0.0025 MAR 0.71386 0.2365 3.018 0.0025 APR 1.4063 0.3496 4.023 0.0001 NOV 7.2145 1.047 6.891 0.0000 DEC 7.5766 1.158 6.542 0.0000 GOA 0.24173 0.7217 0.335 0.7377 TIMELDEP -0.18106 0.2456E-01 -7.371 0.0000 LDEPTH 1.2558 0.2925 4.293 0.0000 L2DEP -0.73829E-01 0.2914E-01 -2.534 0.0113 SLOPE 0.41723E-01 0.4121E-02 10.125 0.0000 SLOPE2 -0.46303E-03 0.1018E-03 -4.548 0.0000 SST 0.47862E-01 0.3521E-01 1.359 0.1741 SST2 0.15012E-02 0.7932E-02 0.189 0.8499 SSTSLOPE 0.19233E-02 0.9740E-03 1.975 0.0483 SSH -0.55611E-01 0.3852E-02 -14.438 0.0000 SSHSLOPE 0.13469E-02 0.6036E-03 2.231 0.0257 LWIND -0.21809 0.1102 -1.979 0.0478 GLDEPTH -0.46192 0.7445E-01 -6.205 0.0000 GTIMELDE 0.80349E-01 0.1155E-01 6.956 0.0000 GSST -0.48696E-01 0.7236E-01 -0.673 0.5010 GSSH 0.59079E-01 0.1065E-01 5.548 0.0000 GLWIND 0.47128 0.2537 1.858 0.0632 IMR3 -0.27612E-01 0.3475E-01 -0.795 0.4268  -0.1202811E+04 0.9037674E+00 Mean Square 0.8791520E+01 0.1423917E+00 0.2682967E+00 Var. Mean  Var. st. dev.  0.1467 0.3302 0.3017 0.0942 0.0335 0.1569 17.8641 4.5226 20.7017 3.6965 86.2419 3.1015 11.2162 12.9840 -5.4090 25.7042 2.3851 0.7443 2.1758 0.7697 -0.5913 0.3706 1.8369  0.3540 0.4704 0.4591 0.2923 0.1801 0.3638 12.1146 0.4979 4.8728 8.5220 319.1512 1.2961 7.6731 11.0375 5.1606 17.0200 0.1233 1.7397 5.5440 1.7962 2.3521 0.8612 0.4658  Economic Valuation Of Critical Habitat Closures, Berman et al.  70  Atka mackerel, standard CPUE (cont.) Limited Dependent Variable Model - CENSORED Maximum Likelihood Estimates Log-Likelihood.............. -301.66 Threshold values for the model: Lower N(0,1) used for significance levels. Variable Coefficient Std. Error Constant -34.660 11.60 FEB 3.8767 1.453 MAR 7.4859 2.822 APR 11.374 4.143 NOV 38.969 26.25 DEC 47.706 13.72 GOA -8.4889 9.698 TIMELDEP -1.0691 0.2922 LDEPTH 14.818 4.787 L2DEP -1.2947 0.4866 SLOPE 0.17656 0.2213E-01 SLOPE2 -0.25598E-02 0.4588E-03 SST 0.21850 0.2181 SST2 0.31178E-01 0.5390E-01 SSTSLOPE 0.20967E-01 0.6283E-02 SSH -0.14980 0.2583E-01 SSHSLOPE 0.84141E-02 0.4288E-02 LWIND -1.7169 0.5096 GLDEPTH -0.73699 0.9166 GTIMELDE 0.72536E-01 0.1191 GSST -2.0576 0.6663 GSSH 0.39086E-01 0.1059 GLWIND 8.2018 3.358 IMR3 0.14239 0.2502 Sigma 1.2081 0.8352E-01  regression 0.0000  Upper  **********  t-ratio -2.988 2.669 2.653 2.745 1.485 3.477 -0.875 -3.659 3.095 -2.661 7.978 -5.580 1.002 0.578 3.337 -5.799 1.962 -3.369 -0.804 0.609 -3.088 0.369 2.442 0.569 14.466  Prob. 0.0028 0.0076 0.0080 0.0060 0.1376 0.0005 0.3814 0.0003 0.0020 0.0078 0.0000 0.0000 0.3165 0.5630 0.0008 0.0000 0.0497 0.0008 0.4214 0.5426 0.0020 0.7120 0.0146 0.5692 0.0000  Var. Mean  Var. st. dev.  0.1467 0.3302 0.3017 0.0942 0.0335 0.1569 17.8641 4.5226 20.7017 3.6965 86.2419 3.1015 11.2162 12.9840 -5.4090 25.7042 2.3851 0.7443 2.1758 0.7697 -0.5913 0.3706 1.8369  0.3540 0.4704 0.4591 0.2923 0.1801 0.3638 12.1146 0.4979 4.8728 8.5220 319.1512 1.2961 7.6731 11.0375 5.1606 17.0200 0.1233 1.7397 5.5440 1.7962 2.3521 0.8612 0.4658  Economic Valuation Of Critical Habitat Closures, Berman et al.  71  Atka mackerel, standard CPUE (cont.) ML Estimates of Selection Model Maximum Likelihood Estimates Log-Likelihood.............. -3700.3 LHS is CENSORED. Tobit Model fit by MLE. FIRST 25 estimates are probit equation. N(0,1) used for significance levels. Variable Coefficient Std. Error Constant -25.650 16.27 FEB 2.5945 1.996 MAR 5.3704 3.816 APR 7.8817 5.663 NOV 30.922 4108. DEC 38.664 18.61 GOA 4.7623 6.405 TIMELDEP -0.86924 0.3946 LDEPTH 11.894 6.859 L2DEP -1.0056 0.7298 SLOPE 0.17329 0.2808E-01 SLOPE2 -0.29793E-02 0.5322E-03 SST 0.23424 0.1940 SST2 -0.19454E-01 0.5202E-01 SSTSLOPE 0.89657E-02 0.5497E-02 SSH -0.11840 0.2624E-01 SSHSLOPE 0.11803E-01 0.5106E-02 LWIND -1.8455 0.5698 GLDEPTH -0.77086 1.749 GTIMELDE 0.55854E-01 0.1737 GSST -1.3869 0.8370 GSSH 0.35503E-01 0.1585 GLWIND 1.8487 0.5711 SIGMA(1) 1.5309 0.1964 RHO(1,2) -0.64896 0.1448  t-ratio -1.577 1.300 1.407 1.392 0.008 2.078 0.743 -2.203 1.734 -1.378 6.172 -5.598 1.207 -0.374 1.631 -4.513 2.312 -3.239 -0.441 0.321 -1.657 0.224 3.237 7.796 -4.482  Prob. 0.1149 0.1936 0.1594 0.1640 0.9940 0.0377 0.4572 0.0276 0.0829 0.1682 0.0000 0.0000 0.2273 0.7084 0.1029 0.0000 0.0208 0.0012 0.6594 0.7478 0.0975 0.8227 0.0012 0.0000 0.0000  Var. Mean  Var. st. dev.  72  Economic Valuation Of Critical Habitat Closures, Berman et al.  Black cod, standard CPUE Sample Selection Model Two stage least squares regression. Dep. Variable LOGBCODS Observations 1653 Weights ONE Mean of LHS 0.1877368E-01 Std.Dev of LHS 0.1331380E+00 StdDev of resid. 0.1099445E+00 Sum of squares 0.1960639E+02 R-squared 0.3176522E+00 Adj. R-squared 0.3050317E+00 F[ 30, 1622] 0.2516956E+02 Prob value 0.3217295E-13 Log-likelihood 0.1319602E+04 Restr.(b=0) Log-l 0.9880533E+03 Amemiya Pr. Criter. 0.1231448E-01 Akaike Info.Crit. -0.1559107E+01 Standard error corrected for selection..... 0.11313 Correlation of disturbance in regression and Selection Criterion (Rho).............. -0.28055 N(0,1) used for significance levels. Variable Coefficient Std. Error t-ratio Prob. Var. Mean Var. st. dev. Constant 2.4274 0.2362 10.276 0.0000 FEB 0.79184E-02 0.3793E-01 0.209 0.8346 0.1506 0.3578 MAR 0.53221E-01 0.7198E-01 0.739 0.4597 0.3315 0.4709 APR 0.61085E-01 0.1059 0.577 0.5639 0.2989 0.4579 NOV 0.35221 0.3101 1.136 0.2560 0.0901 0.2865 DEC 0.39322 0.3375 1.165 0.2440 0.0321 0.1762 GOA 0.78183E-01 0.2188 0.357 0.7208 0.1688 0.3747 TIMELDEP -0.80420E-02 0.7177E-02 -1.121 0.2625 17.6167 11.9465 LDEPTH -1.0480 0.8925E-01 -11.742 0.0000 4.5252 0.4984 L2DEP 0.12000 0.8940E-02 13.423 0.0000 20.7261 4.8625 SLOPE 0.19162E-03 0.1173E-02 0.163 0.8703 3.9975 8.9265 SLOPE2 -0.15904E-04 0.2722E-04 -0.584 0.5591 95.6138 342.6782 SST -0.62122E-02 0.9342E-02 -0.665 0.5061 3.1065 1.3323 SST2 0.24332E-02 0.2249E-02 1.082 0.2792 11.2998 7.7299 SSTSLOPE 0.78424E-04 0.2648E-03 0.296 0.7671 13.1123 11.4851 SSH -0.12560E-03 0.1084E-02 -0.116 0.9078 -5.3569 5.1763 SSHSLOPE 0.22311E-03 0.1731E-03 1.289 0.1974 26.4713 17.4774 MWIND -0.29226E-01 0.3219E-01 -0.908 0.3639 2.3861 0.1216 MCHLA 0.27555E-01 0.1913E-01 1.441 0.1497 0.4752 0.3091 MCHLA1 -0.29923E-01 0.1915E-01 -1.562 0.1182 0.4445 0.3390 DWIND -0.21355E-01 0.1412E-01 -1.513 0.1303 0.0490 0.2159 DCHLA -0.62132E-02 0.1338E-01 -0.464 0.6424 0.2390 0.4266 DCHLA1 -0.65625E-02 0.1078E-01 -0.609 0.5427 0.3696 0.4829 GLDEPTH 0.32166E-01 0.2342E-01 1.374 0.1696 0.8006 1.7913 GTIMELDE 0.14320E-01 0.4161E-02 3.442 0.0006 2.3168 5.6658 GSST -0.81180E-01 0.2034E-01 -3.992 0.0001 0.8209 1.8344 GSSH 0.65528E-02 0.2737E-02 2.394 0.0167 -0.6570 2.5355 GMWIND 0.40875E-01 0.7596E-01 0.538 0.5905 0.3997 0.8888 GMCHLA -0.70419E-01 0.2949E-01 -2.388 0.0170 0.0888 0.2504 GMCHLA1 -0.18011 0.8041E-01 -2.240 0.0251 0.0497 0.1298 IMR2 -0.31739E-01 0.1040E-01 -3.053 0.0023 1.1205 0.4131  Economic Valuation Of Critical Habitat Closures, Berman et al.  73  Black cod, standard CPUE (cont.) Limited Dependent Variable Model - CENSORED Maximum Likelihood Estimates Log-Likelihood.............. -220.25 Threshold values for the model: Lower N(0,1) used for significance levels. Variable Coefficient Std. Error Constant 3.8818 1.656 FEB 0.10332 0.2868 MAR -0.22769E-01 0.5783 APR -0.18441 0.8442 NOV -2.5711 7.886 DEC -2.2801 16.09 GOA -0.62586 1.308 TIMELDEP 0.43199E-01 0.5101E-01 LDEPTH -1.9790 0.6414 L2DEP 0.24043 0.6431E-01 SLOPE -0.26325E-02 0.1051E-01 SLOPE2 -0.24089E-03 0.3034E-03 SST -0.99882E-01 0.7856E-01 SST2 0.25173E-01 0.1688E-01 SSTSLOPE -0.45292E-03 0.2235E-02 SSH 0.33618E-01 0.1035E-01 SSHSLOPE 0.29973E-02 0.1254E-02 MWIND -0.14052 0.2238 MCHLA 0.26636 0.2143 MCHLA1 -0.70390 0.3591 DWIND -0.54923 0.2197 DCHLA 0.11690 0.9831E-01 DCHLA1 -0.29439E-01 0.1487 GLDEPTH 0.23380 0.1471 GTIMELDE -0.64460E-02 0.2493E-01 GSST -0.21247 0.1357 GSSH -0.52664E-01 0.1856E-01 GMWIND 0.23113 0.4277 GMCHLA -0.84100E-01 0.2490 GMCHLA1 -0.17378 0.6210 IMR2 -0.28799 0.8733E-01 Sigma 0.37722 0.2295E-01  regression 0.0000  Upper  **********  t-ratio 2.345 0.360 -0.039 -0.218 -0.326 -0.142 -0.478 0.847 -3.085 3.739 -0.251 -0.794 -1.271 1.491 -0.203 3.248 2.391 -0.628 1.243 -1.960 -2.500 1.189 -0.198 1.589 -0.259 -1.566 -2.838 0.540 -0.338 -0.280 -3.298 16.440  Prob. 0.0190 0.7187 0.9686 0.8271 0.7444 0.8873 0.6323 0.3970 0.0020 0.0002 0.8022 0.4272 0.2036 0.1359 0.8394 0.0012 0.0168 0.5301 0.2140 0.0500 0.0124 0.2344 0.8431 0.1120 0.7959 0.1174 0.0045 0.5889 0.7356 0.7796 0.0010 0.0000  Var. Mean  Var. st. dev.  0.1514 0.3329 0.2969 0.0895 0.0319 0.1683 17.5742 4.5267 20.7384 4.0268 96.6545 3.1098 11.3145 13.0559 -5.3600 26.4905 2.3863 0.4744 0.4432 0.0487 0.2386 0.3720 0.7980 2.3070 0.8186 -0.6535 0.3985 0.0885 0.0494 1.8280  0.3586 0.4714 0.4570 0.2856 0.1757 0.3742 11.9220 0.4973 4.8504 8.9715 345.2532 1.3297 7.7193 11.4707 5.1714 17.4916 0.1215 0.3084 0.3385 0.2153 0.4263 0.4835 1.7888 5.6523 1.8326 2.5277 0.8879 0.2498 0.1294 0.4850  Economic Valuation Of Critical Habitat Closures, Berman et al.  74  Black cod, standard CPUE (cont.) ML Estimates of Selection Model Maximum Likelihood Estimates Log-Likelihood.............. -3543.2 LHS is CENSORED. Tobit Model fit by MLE. FIRST 30 estimates are probit equation. N(0,1) used for significance levels. Variable Coefficient Std. Error Constant 2.9620 2.057 FEB 0.11760 0.3187 MAR 0.42049E-03 0.6920 APR -0.19232 0.9516 NOV -2.4524 1232. DEC -1.6980 327.3 GOA -0.37774 1.709 TIMELDEP 0.38390E-01 0.4995E-01 LDEPTH -1.5918 0.7620 L2DEP 0.19778 0.7363E-01 SLOPE -0.51571E-02 0.1535E-01 SLOPE2 -0.10672E-03 0.4393E-03 SST -0.77259E-01 0.1102 SST2 0.25042E-01 0.2403E-01 SSTSLOPE -0.23213E-03 0.2909E-02 SSH 0.34729E-01 0.1487E-01 SSHSLOPE 0.29496E-02 0.1493E-02 MWIND -0.23381 0.3027 MCHLA 0.22720 0.3550 MCHLA1 -0.64249 0.6993 DWIND -0.55175 0.3753 DCHLA 0.60389E-01 0.1296 DCHLA1 -0.10837 0.2019 GLDEPTH 0.22527 0.1480 GTIMELDE -0.80087E-02 0.2776E-01 GSST -0.26293 0.1854 GSSH -0.54842E-01 0.2298E-01 GMWIND 0.21476 0.5189 GMCHLA -0.12324 0.4085 GMCHLA1 -0.31465 0.9677 SIGMA(1) 0.41090 0.4339E-01 RHO(1,2) -0.41769 0.2496  t-ratio 1.440 0.369 0.001 -0.202 -0.002 -0.005 -0.221 0.769 -2.089 2.686 -0.336 -0.243 -0.701 1.042 -0.080 2.336 1.975 -0.773 0.640 -0.919 -1.470 0.466 -0.537 1.522 -0.289 -1.418 -2.386 0.414 -0.302 -0.325 9.471 -1.674  Prob. 0.1499 0.7121 0.9995 0.8398 0.9984 0.9959 0.8251 0.4422 0.0367 0.0072 0.7368 0.8080 0.4833 0.2973 0.9364 0.0195 0.0482 0.4398 0.5222 0.3582 0.1415 0.6413 0.5914 0.1281 0.7729 0.1562 0.0170 0.6789 0.7629 0.7451 0.0000 0.0942  Var. Mean  Var. st. dev.  Economic Valuation Of Critical Habitat Closures, Berman et al.  75  Rockfish, standard CPUE Sample Selection Model Two stage least squares regression. Dep. Variable LOGROCKS Observations 1653 Weights ONE Mean of LHS 0.9454465E-01 Std.Dev of LHS 0.3560142E+00 StdDev of resid. 0.2955764E+00 Sum of squares 0.1417067E+03 R-squared 0.3102881E+00 Adj. R-squared 0.2975315E+00 F[ 30, 1622] 0.2432356E+02 Prob value 0.3217295E-13 Log-likelihood -0.3151357E+03 Restr.(b=0) Log-l Amemiya Pr. Criter. 0.8900384E-01 Akaike Info.Crit. Standard error corrected for selection..... 0.29652 Correlation of disturbance in regression and Selection Criterion (Rho).............. -0.94950E-01 N(0,1) used for significance levels. Variable Coefficient Std. Error t-ratio Prob. Constant 0.85655 0.6326 1.354 0.1757 FEB 0.83449E-02 0.1019 0.082 0.9348 MAR -0.53336E-01 0.1937 -0.275 0.7830 APR -0.29167E-01 0.2848 -0.102 0.9184 NOV 0.81535 0.8349 0.977 0.3288 DEC 0.82764 0.9089 0.911 0.3625 GOA -0.88707 0.5878 -1.509 0.1313 TIMELDEP -0.21835E-01 0.1932E-01 -1.130 0.2585 LDEPTH -0.17497 0.2391 -0.732 0.4643 L2DEP 0.31109E-01 0.2394E-01 1.299 0.1938 SLOPE 0.36580E-01 0.3118E-02 11.732 0.0000 SLOPE2 -0.61869E-03 0.7224E-04 -8.565 0.0000 SST -0.11381E-01 0.2497E-01 -0.456 0.6486 SST2 0.51224E-02 0.6026E-02 0.850 0.3953 SSTSLOPE -0.60695E-03 0.7056E-03 -0.860 0.3897 SSH -0.19526E-01 0.2901E-02 -6.731 0.0000 SSHSLOPE 0.90913E-03 0.4612E-03 1.971 0.0487 MWIND -0.21572 0.8579E-01 -2.514 0.0119 MCHLA 0.32825E-01 0.5114E-01 0.642 0.5210 MCHLA1 -0.68217E-02 0.5122E-01 -0.133 0.8941 DWIND -0.66462E-01 0.3771E-01 -1.763 0.0780 DCHLA 0.29790E-01 0.3571E-01 0.834 0.4042 DCHLA1 -0.17345E-01 0.2874E-01 -0.603 0.5462 GLDEPTH -0.29733E-02 0.6303E-01 -0.047 0.9624 GTIMELDE 0.20665E-01 0.1119E-01 1.847 0.0648 GSST 0.66855E-01 0.5470E-01 1.222 0.2217 GSSH 0.16746E-01 0.7368E-02 2.273 0.0230 GMWIND 0.17361 0.2040 0.851 0.3947 GMCHLA -0.27873E-01 0.7928E-01 -0.352 0.7252 GMCHLA1 -0.11691 0.2161 -0.541 0.5885 IMR2 -0.28155E-01 0.2768E-01 -1.017 0.3091  -0.6378122E+03 0.4187970E+00  Var. Mean  Var. st. dev.  0.1506 0.3315 0.2989 0.0901 0.0321 0.1688 17.6167 4.5252 20.7261 3.9975 95.6138 3.1065 11.2998 13.1123 -5.3569 26.4713 2.3861 0.4752 0.4445 0.0490 0.2390 0.3696 0.8006 2.3168 0.8209 -0.6570 0.3997 0.0888 0.0497 1.1206  0.3578 0.4709 0.4579 0.2865 0.1762 0.3747 11.9465 0.4984 4.8625 8.9265 342.6782 1.3323 7.7299 11.4851 5.1763 17.4774 0.1216 0.3091 0.3390 0.2159 0.4266 0.4829 1.7913 5.6658 1.8344 2.5355 0.8888 0.2504 0.1298 0.4129  Economic Valuation Of Critical Habitat Closures, Berman et al.  76  Rockfish, standard CPUE (cont.) Limited Dependent Variable Model - CENSORED Maximum Likelihood Estimates Log-Likelihood.............. -513.72 Threshold values for the model: Lower N(0,1) used for significance levels. Variable Coefficient Std. Error Constant -11.694 3.018 FEB -0.41568 0.5026 MAR -1.2020 0.9679 APR -2.1255 1.460 NOV -5.9906 15.44 DEC -7.1441 28.30 GOA 2.3880 1.931 TIMELDEP 0.11025 0.9308E-01 LDEPTH 4.4729 1.162 L2DEP -0.38035 0.1172 SLOPE 0.77617E-01 0.8660E-02 SLOPE2 -0.12622E-02 0.1915E-03 SST -0.22672 0.8609E-01 SST2 0.60631E-01 0.2358E-01 SSTSLOPE -0.35217E-02 0.2518E-02 SSH -0.67012E-01 0.1124E-01 SSHSLOPE 0.18710E-02 0.1694E-02 MWIND -0.96248 0.3144 MCHLA -0.39986 0.4040 MCHLA1 -0.73409 0.6102 DWIND 0.11798E-01 0.1161 DCHLA 0.24601 0.1559 DCHLA1 0.34102E-01 0.1879 GLDEPTH -1.0949 0.2173 GTIMELDE 0.97137E-01 0.3384E-01 GSST -0.11523 0.1893 GSSH 0.77479E-01 0.2297E-01 GMWIND 1.0618 0.6392 GMCHLA 0.55717 0.4411 GMCHLA1 -0.25825 0.7854 IMR2 -0.39212E-01 0.9918E-01 Sigma 0.67367 0.2677E-01  regression 0.0000  Upper  **********  t-ratio -3.875 -0.827 -1.242 -1.456 -0.388 -0.252 1.237 1.184 3.850 -3.246 8.963 -6.592 -2.634 2.572 -1.399 -5.961 1.104 -3.061 -0.990 -1.203 0.102 1.578 0.182 -5.038 2.871 -0.609 3.373 1.661 1.263 -0.329 -0.395 25.163  Prob. 0.0001 0.4082 0.2143 0.1455 0.6979 0.8007 0.2163 0.2362 0.0001 0.0012 0.0000 0.0000 0.0084 0.0101 0.1619 0.0000 0.2694 0.0022 0.3223 0.2289 0.9191 0.1145 0.8560 0.0000 0.0041 0.5426 0.0007 0.0967 0.2066 0.7423 0.6926 0.0000  Var. Mean  Var. st. dev.  0.1510 0.3339 0.2972 0.0897 0.0319 0.1685 17.5917 4.5260 20.7319 3.9858 95.1317 3.1086 11.3079 13.0662 -5.3578 26.4753 2.3861 0.4748 0.4438 0.0487 0.2377 0.3706 0.7990 2.3097 0.8195 -0.6543 0.3990 0.0886 0.0495 1.8279  0.3582 0.4718 0.4572 0.2858 0.1758 0.3744 11.9199 0.4973 4.8508 8.9047 341.8117 1.3300 7.7204 11.4736 5.1643 17.4747 0.1214 0.3084 0.3384 0.2154 0.4258 0.4831 1.7896 5.6551 1.8335 2.5291 0.8884 0.2500 0.1295 0.4850  Economic Valuation Of Critical Habitat Closures, Berman et al.  77  Rockfish, standard CPUE (cont.) ML Estimates of Selection Model Maximum Likelihood Estimates Log-Likelihood.............. -3838.3 LHS is CENSORED. Tobit Model fit by MLE. FIRST 30 estimates are probit equation. N(0,1) used for significance levels. Variable Coefficient Std. Error t-ratio ---------------------------------------------------------------------------Constant -11.433 3.360 -3.403 FEB -0.44676 0.5944 -0.752 MAR -1.1717 1.155 -1.015 APR -2.1339 1.771 -1.205 NOV -5.5952 487.8 -0.011 DEC -6.9311 3230. -0.002 GOA 2.1761 2.591 0.840 TIMELDEP 0.11665 0.1176 0.992 LDEPTH 4.3433 1.333 3.259 L2DEP -0.36760 0.1366 -2.692 SLOPE 0.74790E-01 0.9806E-02 7.627 SLOPE2 -0.12404E-02 0.2055E-03 -6.037 SST -0.22718 0.1004 -2.264 SST2 0.60625E-01 0.2731E-01 2.220 SSTSLOPE -0.37243E-02 0.3075E-02 -1.211 SSH -0.67868E-01 0.1097E-01 -6.185 SSHSLOPE 0.18257E-02 0.1723E-02 1.059 MWIND -0.93293 0.2593 -3.598 MCHLA -0.42820 0.3984 -1.075 MCHLA1 -0.85148 0.8477 -1.004 DWIND 0.24351E-01 0.1556 0.157 DCHLA 0.25822 0.1700 1.519 DCHLA1 0.97265E-01 0.2350 0.414 GLDEPTH -1.0709 0.2429 -4.409 GTIMELDE 0.95116E-01 0.4150E-01 2.292 GSST -0.94056E-01 0.2143 -0.439 GSSH 0.77905E-01 0.2516E-01 3.096 GMWIND 1.0637 0.8297 1.282 GMCHLA 0.59288 0.4589 1.292 GMCHLA1 -0.22814 1.057 -0.216 SIGMA(1) 0.68102 0.2746E-01 24.803 RHO(1,2) -0.17018 0.1592 -1.069  Prob. 0.0007 0.4523 0.3103 0.2283 0.9908 0.9983 0.4011 0.3214 0.0011 0.0071 0.0000 0.0000 0.0236 0.0264 0.2258 0.0000 0.2894 0.0003 0.2824 0.3151 0.8756 0.1288 0.6790 0.0000 0.0219 0.6607 0.0020 0.1998 0.1964 0.8292 0.0000 0.2851  78  Economic Valuation Of Critical Habitat Closures, Berman et al.  Flatfish, standard CPUE Sample Selection Model Two stage least squares regression. Dep. Variable LOGFLATS Observations 1653 Weights ONE Mean of LHS 0.1157577E+01 Std.Dev of LHS 0.8117898E+00 StdDev of resid. 0.6808442E+00 Sum of squares R-squared 0.2961645E+00 Adj. R-squared 0.2831466E+00 F[ 30, 1622] 0.2275053E+02 Prob value 0.3217295E-13 Log-likelihood -0.1694409E+04 Restr.(b=0) Log-l Amemiya Pr. Criter. 0.4722421E+00 Akaike Info.Crit. Standard error corrected for selection..... 0.68107 Correlation of disturbance in regression and Selection Criterion (Rho).............. -0.30324E-01 N(0,1) used for significance levels. Variable Coefficient Std. Error t-ratio Prob. Constant 6.1487 1.457 4.221 0.0000 FEB -0.97080 0.2348 -4.135 0.0000 MAR -2.0991 0.4461 -4.706 0.0000 APR -3.2206 0.6562 -4.908 0.0000 NOV -8.5532 1.924 -4.447 0.0000 DEC -8.8377 2.094 -4.221 0.0000 GOA -3.5439 1.354 -2.618 0.0089 TIMELDEP 0.19577 0.4452E-01 4.397 0.0000 LDEPTH -2.0645 0.5506 -3.750 0.0002 L2DEP 0.21101 0.5513E-01 3.827 0.0001 SLOPE -0.73404E-01 0.7173E-02 -10.233 0.0000 SLOPE2 0.11978E-02 0.1661E-03 7.209 0.0000 SST 0.24649 0.5749E-01 4.288 0.0000 SST2 -0.74073E-01 0.1388E-01 -5.338 0.0000 SSTSLOPE -0.25282E-02 0.1624E-02 -1.557 0.1195 SSH 0.40100E-01 0.6678E-02 6.005 0.0000 SSHSLOPE -0.12754E-02 0.1061E-02 -1.202 0.2295 MWIND -0.25255 0.1974 -1.279 0.2008 MCHLA 1.1488 0.1177 9.757 0.0000 MCHLA1 0.86893E-01 0.1179 0.737 0.4612 DWIND -0.22258 0.8680E-01 -2.564 0.0103 DCHLA -0.13774 0.8220E-01 -1.676 0.0938 DCHLA1 -0.88872E-01 0.6615E-01 -1.344 0.1791 GLDEPTH 0.14648 0.1452 1.009 0.3130 GTIMELDE 0.88725E-02 0.2577E-01 0.344 0.7307 GSST 0.38910 0.1260 3.088 0.0020 GSSH -0.65376E-01 0.1697E-01 -3.852 0.0001 GMWIND 0.50701 0.4698 1.079 0.2805 GMCHLA -1.0678 0.1826 -5.847 0.0000 GMCHLA1 0.56283 0.4977 1.131 0.2581 IMR2 -0.20653E-01 0.6372E-01 -0.324 0.7458  0.7518762E+03 -0.2000332E+04 0.2087609E+01  Var. Mean  Var. st. dev.  0.1506 0.3315 0.2989 0.0901 0.0321 0.1688 17.6167 4.5252 20.7261 3.9975 95.6138 3.1065 11.2998 13.1123 -5.3569 26.4713 2.3861 0.4752 0.4445 0.0490 0.2390 0.3696 0.8006 2.3168 0.8209 -0.6570 0.3997 0.0888 0.0497 1.1206  0.3578 0.4709 0.4579 0.2865 0.1762 0.3747 11.9465 0.4984 4.8625 8.9265 342.6782 1.3323 7.7299 11.4851 5.1763 17.4774 0.1216 0.3091 0.3390 0.2159 0.4266 0.4829 1.7913 5.6658 1.8344 2.5355 0.8888 0.2504 0.1298 0.4129  Economic Valuation Of Critical Habitat Closures, Berman et al.  79  Flatfish, standard CPUE (cont.) Limited Dependent Variable Model - CENSORED Maximum Likelihood Estimates Log-Likelihood.............. -1721.2 Threshold values for the model: Lower N(0,1) used for significance levels. Variable Coefficient Std. Error Constant 6.8473 1.489 FEB -1.0425 0.2388 MAR -2.2391 0.4534 APR -3.4395 0.6642 NOV -9.4973 1.947 DEC -9.8708 2.121 GOA -3.6358 1.371 TIMELDEP 0.22064 0.4513E-01 LDEPTH -2.3806 0.5638 L2DEP 0.24184 0.5660E-01 SLOPE -0.81900E-01 0.7336E-02 SLOPE2 0.12876E-02 0.1759E-03 SST 0.22309 0.5891E-01 SST2 -0.73641E-01 0.1409E-01 SSTSLOPE -0.28536E-02 0.1661E-02 SSH 0.46324E-01 0.6920E-02 SSHSLOPE -0.12551E-02 0.1076E-02 MWIND -0.19251 0.2035 MCHLA 1.1717 0.1205 MCHLA1 0.66035E-01 0.1195 DWIND -0.18149 0.8857E-01 DCHLA -0.11872 0.8521E-01 DCHLA1 -0.48766E-01 0.6881E-01 GLDEPTH 0.16995 0.1475 GTIMELDE 0.51625E-02 0.2618E-01 GSST 0.42715 0.1283 GSSH -0.69744E-01 0.1727E-01 GMWIND 0.47162 0.4768 GMCHLA -1.0817 0.1852 GMCHLA1 0.59237 0.5041 IMR2 -0.10502 0.6598E-01 Sigma 0.68992 0.1216E-01  regression 0.0000  Upper  **********  t-ratio 4.600 -4.366 -4.939 -5.178 -4.878 -4.655 -2.653 4.889 -4.223 4.273 -11.164 7.321 3.787 -5.226 -1.718 6.694 -1.166 -0.946 9.726 0.552 -2.049 -1.393 -0.709 1.152 0.197 3.331 -4.039 0.989 -5.839 1.175 -1.592 56.747  Prob. 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0080 0.0000 0.0000 0.0000 0.0000 0.0000 0.0002 0.0000 0.0858 0.0000 0.2436 0.3442 0.0000 0.5806 0.0404 0.1635 0.4785 0.2492 0.8437 0.0009 0.0001 0.3226 0.0000 0.2400 0.1114 0.0000  Var. Mean  Var. st. dev.  0.1511 0.3311 0.2985 0.0900 0.0320 0.1686 17.6044 4.5258 20.7312 4.0238 96.7721 3.1071 11.3031 13.1029 -5.3603 26.4725 2.3863 0.4749 0.4441 0.0489 0.2393 0.3704 0.7996 2.3140 0.8199 -0.6562 0.3992 0.0887 0.0496 1.8292  0.3582 0.4708 0.4577 0.2863 0.1761 0.3745 11.9448 0.4984 4.8619 8.9794 345.8994 1.3321 7.7297 11.4816 5.1831 17.4673 0.1216 0.3090 0.3390 0.2158 0.4268 0.4831 1.7904 5.6629 1.8335 2.5341 0.8884 0.2503 0.1297 0.4858  Economic Valuation Of Critical Habitat Closures, Berman et al.  80  Flatfish, standard CPUE (cont.) ML Estimates of Selection Model Maximum Likelihood Estimates Log-Likelihood.............. -5040.6 LHS is CENSORED. Tobit Model fit by MLE. FIRST 30 estimates are probit equation. N(0,1) used for significance levels. Variable Coefficient Std. Error Constant 6.1914 1.555 FEB -0.95373 0.2786 MAR -2.0431 0.5603 APR -3.1565 0.8300 NOV -8.6509 2.469 DEC -8.9204 2.686 GOA -3.4433 1.535 TIMELDEP 0.19935 0.5732E-01 LDEPTH -2.1238 0.5971 L2DEP 0.21786 0.6183E-01 SLOPE -0.82142E-01 0.7867E-02 SLOPE2 0.13501E-02 0.1894E-03 SST 0.23645 0.6877E-01 SST2 -0.72843E-01 0.1542E-01 SSTSLOPE -0.26437E-02 0.1914E-02 SSH 0.46902E-01 0.7795E-02 SSHSLOPE -0.14632E-02 0.1132E-02 MWIND -0.23509 0.1815 MCHLA 1.1495 0.1240 MCHLA1 0.85693E-01 0.1634 DWIND -0.20470 0.1068 DCHLA -0.14862 0.9172E-01 DCHLA1 -0.83526E-01 0.6834E-01 GLDEPTH 0.17628 0.1497 GTIMELDE 0.48022E-02 0.2465E-01 GSST 0.37656 0.1239 GSSH -0.70388E-01 0.1600E-01 GMWIND 0.46083 0.5310 GMCHLA -1.0867 0.2070 GMCHLA1 0.50834 0.5185 SIGMA(1) 0.69040 0.1242E-01 RHO(1,2) -0.93186E-02 0.1119  t-ratio 3.981 -3.423 -3.646 -3.803 -3.504 -3.321 -2.244 3.478 -3.557 3.524 -10.441 7.129 3.438 -4.725 -1.381 6.017 -1.293 -1.295 9.271 0.525 -1.917 -1.620 -1.222 1.177 0.195 3.039 -4.401 0.868 -5.250 0.980 55.593 -0.083  Prob. 0.0001 0.0006 0.0003 0.0001 0.0005 0.0009 0.0249 0.0005 0.0004 0.0004 0.0000 0.0000 0.0006 0.0000 0.1672 0.0000 0.1960 0.1952 0.0000 0.5999 0.0552 0.1052 0.2216 0.2391 0.8456 0.0024 0.0000 0.3855 0.0000 0.3269 0.0000 0.9336  Var. Mean  Var. st. dev.  Economic Valuation Of Critical Habitat Closures, Berman et al.  81  2. Summer bottom trawl selection equation (IMR7) Binomial Probit Model Maximum Likelihood Estimates Log-Likelihood.............. -1125.6 Restricted (Slopes 0) Log-L. -5571.7 Chi-Squared (29)............ 8892.3 Significance Level.......... 0.32173E-13 N(0,1) used for significance levels. Variable Coefficient Std. Error Constant -8.5656 1.161 JUN -0.56585 0.1549 JUL 0.14640 0.1579 AUG 0.25194 0.1892 SEP 0.64802E-01 0.1891 OCT 0.32963 0.1780 GOA -0.58712 0.2235 LDEPTH 3.5841 0.4522 L2DEP -0.40491 0.4853E-01 SLOPE 0.27611E-01 0.1268E-01 SLOPE2 0.13874E-03 0.3151E-03 SST 0.14330 0.5170E-01 SST2 -0.80494E-02 0.3063E-02 SSTSLOPE -0.12476E-02 0.1904E-02 SSH -0.52594E-02 0.9747E-02 SSHSLOPE -0.25610E-02 0.1810E-02 MWIND -0.55131 0.2237 MCHLA 0.18774 0.6358E-01 MCHLA1 0.18947 0.5607E-01 DWIND -0.46088 0.1402 DCHLA -0.19016 0.1275 DCHLA1 -0.25206 0.1414 POLTRAWL -0.41931 0.1364 CODTRAWL 0.34204 0.2689 ATKTRAWL 1.0861 0.3081 POLTSSL -0.68190 0.2108 CODTSSL 0.21356 0.2041 ATKTSSL 0.30112 0.1609 MIXTSSL -1.7791 0.3092 PORTDIST -0.13200E-01 0.1235E-02 Frequencies of actual & predicted outcomes Predicted outcome has maximum probability.  t-ratio -7.379 -3.652 0.927 1.331 0.343 1.852 -2.627 7.927 -8.344 2.178 0.440 2.772 -2.628 -0.655 -0.540 -1.415 -2.464 2.953 3.379 -3.288 -1.491 -1.783 -3.074 1.272 3.525 -3.235 1.046 1.872 -5.755 -10.686  Predicted Actual  0  1  TOTAL  0 1  9883 2096  8 9  9891 2105  Total  11979  17  11996  Prob. 0.0000 0.0003 0.3538 0.1831 0.7318 0.0641 0.0086 0.0000 0.0000 0.0294 0.6597 0.0056 0.0086 0.5123 0.5895 0.1571 0.0137 0.0031 0.0007 0.0010 0.1360 0.0746 0.0021 0.2034 0.0004 0.0012 0.2955 0.0613 0.0000 0.0000  Var. Mean  Var. st. dev.  0.1624 0.1657 0.1720 0.1658 0.1670 0.2506 4.4273 20.5808 3.8647 81.2443 6.9487 60.6119 21.5992 -3.5008 27.8622 2.0435 1.0271 1.0011 0.1624 0.0892 0.1103 0.6967 0.7855 0.0105 0.1797 0.1739 0.1545 0.0891 66.4344  0.3689 0.3718 0.3774 0.3719 0.3730 0.4334 0.9897 9.0589 8.1433 275.9448 3.5439 47.4101 16.9362 4.7903 18.4854 0.2532 0.5401 0.5225 0.3689 0.2850 0.3133 0.4456 0.3899 0.0962 0.3654 0.3605 0.3416 0.2847 40.5635  82  Economic Valuation Of Critical Habitat Closures, Berman et al.  Pollock , standard CPUE Sample Selection Model Two stage least squares regression. Dep. Variable LOGPOLLS Observations 2084 Weights ONE Mean of LHS 0.5367812E+00 Std.Dev of LHS 0.6599354E+00 StdDev of resid. 0.5753294E+00 Sum of squares 0.6795509E+03 R-squared 0.2396060E+00 Adj. R-squared F[ 30, 2053] 0.2156386E+02 Prob value 0.3217295E-13 Log-likelihood -0.1789390E+04 Restr.(b=0) Log-l Amemiya Pr. Criter. 0.3359276E+00 Akaike Info.Crit. Standard error corrected for selection..... 0.57811 Correlation of disturbance in regression and Selection Criterion (Rho).............. -0.10700 N(0,1) used for significance levels. Variable Coefficient Std. Error t-ratio Prob. Constant -4.9508 0.8283 -5.977 0.0000 JUN -0.14520 0.1470 -0.988 0.3232 JUL -0.33695 0.2141 -1.574 0.1155 AUG -0.64359 0.3005 -2.141 0.0322 SEP -1.0007 0.3844 -2.603 0.0092 OCT -1.4103 0.4666 -3.022 0.0025 GOA -3.6998 0.7372 -5.019 0.0000 TIMELDEP 0.53041E-01 0.2063E-01 2.571 0.0101 LDEPTH 2.0966 0.3288 6.376 0.0000 L2DEP -0.25616 0.3397E-01 -7.541 0.0000 SLOPE -0.27787E-01 0.5262E-02 -5.280 0.0000 SLOPE2 0.31872E-03 0.1174E-03 2.715 0.0066 SST 0.10120 0.3953E-01 2.560 0.0105 SST2 -0.57602E-02 0.2925E-02 -1.969 0.0489 SSTSLOPE -0.13557E-02 0.9203E-03 -1.473 0.1407 SSH -0.29468E-02 0.6169E-02 -0.478 0.6329 SSHSLOPE 0.20828E-02 0.8658E-03 2.406 0.0161 MWIND 0.34110E-01 0.1392 0.245 0.8065 MCHLA -0.70842E-01 0.4116E-01 -1.721 0.0852 MCHLA1 0.42190E-02 0.2952E-01 0.143 0.8864 DWIND 0.88128E-01 0.8618E-01 1.023 0.3065 DCHLA 0.18086 0.6690E-01 2.703 0.0069 DCHLA1 -0.18743 0.7207E-01 -2.601 0.0093 GLDEPTH 0.34532 0.8189E-01 4.217 0.0000 GTIMELDE 0.11532E-01 0.8576E-02 1.345 0.1787 GSST -0.16840E-01 0.2986E-01 -0.564 0.5728 GSSH -0.19891E-02 0.1075E-01 -0.185 0.8532 GMWIND 0.42804 0.2508 1.707 0.0879 GMCHLA 0.27099 0.8707E-01 3.112 0.0019 GMCHLA1 0.80668E-01 0.8036E-01 1.004 0.3155 IMR7 -0.61856E-01 0.4819E-01 -1.284 0.1993  0.2284945E+00 -0.2090430E+04 0.1747015E+01  Var. Mean  Var. st. dev.  0.0614 0.2692 0.2116 0.1569 0.1934 0.2404 35.5014 4.5578 21.1029 4.3812 106.1247 8.1178 71.7304 20.1822 -3.7809 22.9680 2.0168 1.0076 1.0345 0.0288 0.0509 0.0398 1.1783 8.5557 2.3308 -0.8258 0.4961 0.2574 0.2479 1.9159  0.2402 0.4436 0.4085 0.3638 0.3950 0.4274 7.3817 0.5740 5.4919 9.3259 364.6537 2.4189 37.6320 15.0117 3.8750 16.3217 0.2648 0.4184 0.5422 0.1673 0.2198 0.1956 2.1088 15.7173 4.3117 2.2383 0.8907 0.4944 0.4876 0.4657  Economic Valuation Of Critical Habitat Closures, Berman et al.  83  Pollock , standard CPUE (cont.) Limited Dependent Variable Model - CENSORED Maximum Likelihood Estimates Log-Likelihood.............. -1922.1 Threshold values for the model: Lower N(0,1) used for significance levels. Variable Coefficient Std. Error Constant -7.8763 1.039 JUN -0.12605 0.1785 JUL -0.29978 0.2642 AUG -0.57325 0.3724 SEP -0.89335 0.4775 OCT -1.2361 0.5789 GOA -6.5638 0.9770 TIMELDEP 0.48001E-01 0.2536E-01 LDEPTH 3.4986 0.4334 L2DEP -0.39135 0.4309E-01 SLOPE -0.38735E-01 0.6325E-02 SLOPE2 0.45389E-03 0.1403E-03 SST 0.38036E-01 0.4857E-01 SST2 -0.11547E-02 0.3548E-02 SSTSLOPE -0.13083E-02 0.1065E-02 SSH -0.45853E-02 0.7180E-02 SSHSLOPE 0.16441E-02 0.1022E-02 MWIND -0.79378E-01 0.1617 MCHLA -0.43299E-01 0.4703E-01 MCHLA1 -0.31080E-02 0.3347E-01 DWIND -0.40418E-01 0.1086 DCHLA 0.21986 0.7699E-01 DCHLA1 -0.19798 0.8351E-01 GLDEPTH 0.63679 0.1047 GTIMELDE 0.11580E-01 0.1060E-01 GSST -0.33178E-01 0.3736E-01 GSSH -0.18198E-01 0.1383E-01 GMWIND 0.84511 0.3283 GMCHLA 0.48930 0.1085 GMCHLA1 0.19610 0.9800E-01 IMR7 -0.11926 0.5738E-01 Sigma 0.64576 0.1118E-01  regression 0.0000  Upper  **********  t-ratio -7.581 -0.706 -1.135 -1.539 -1.871 -2.135 -6.719 1.893 8.072 -9.083 -6.125 3.234 0.783 -0.325 -1.229 -0.639 1.609 -0.491 -0.921 -0.093 -0.372 2.856 -2.371 6.085 1.092 -0.888 -1.316 2.574 4.508 2.001 -2.078 57.744  Prob. 0.0000 0.4800 0.2566 0.1238 0.0614 0.0327 0.0000 0.0583 0.0000 0.0000 0.0000 0.0012 0.4335 0.7448 0.2191 0.5231 0.1077 0.6235 0.3572 0.9260 0.7098 0.0043 0.0177 0.0000 0.2748 0.3745 0.1881 0.0100 0.0000 0.0454 0.0377 0.0000  Var. Mean  Var. st. dev.  0.0612 0.2697 0.2109 0.1573 0.1937 0.2425 35.5104 4.5577 21.1014 4.4027 106.8498 8.1238 71.8300 20.1950 -3.7775 23.0271 2.0169 1.0078 1.0345 0.0287 0.0507 0.0397 1.1873 8.6344 2.3526 -0.8332 0.5004 0.2599 0.2504 1.9109  0.2398 0.4439 0.4080 0.3642 0.3953 0.4287 7.3774 0.5732 5.4836 9.3546 365.4387 2.4194 37.6711 15.0092 3.8726 16.3453 0.2647 0.4181 0.5419 0.1670 0.2194 0.1953 2.1130 15.7704 4.3267 2.2430 0.8934 0.4961 0.4896 0.4671  Economic Valuation Of Critical Habitat Closures, Berman et al.  84  Pollock , standard CPUE (cont.) ML Estimates of Selection Model Maximum Likelihood Estimates Log-Likelihood.............. -6240.2 LHS is CENSORED. Tobit Model fit by MLE. FIRST 30 estimates are probit equation. N(0,1) used for significance levels. Variable Coefficient Std. Error Constant -8.8557 1.161 JUN -0.21361 0.2191 JUL -0.34851 0.3123 AUG -0.64471 0.4357 SEP -0.97808 0.5669 OCT -1.3192 0.6900 GOA -6.4392 1.170 TIMELDEP 0.55940E-01 0.2903E-01 LDEPTH 3.8321 0.4988 L2DEP -0.43531 0.4700E-01 SLOPE -0.34657E-01 0.7108E-02 SLOPE2 0.42922E-03 0.1748E-03 SST 0.53129E-01 0.6353E-01 SST2 -0.17815E-02 0.4292E-02 SSTSLOPE -0.10514E-02 0.1149E-02 SSH -0.48489E-02 0.6528E-02 SSHSLOPE 0.15819E-02 0.1004E-02 MWIND -0.19282 0.1593 MCHLA -0.27358E-02 0.5160E-01 MCHLA1 0.13265E-01 0.3232E-01 DWIND -0.10874 0.9448E-01 DCHLA 0.17846 0.7966E-01 DCHLA1 -0.22222 0.7962E-01 GLDEPTH 0.62990 0.1295 GTIMELDE 0.84452E-02 0.1222E-01 GSST -0.36456E-01 0.4259E-01 GSSH -0.19688E-01 0.1557E-01 GMWIND 0.86262 0.3922 GMCHLA 0.47273 0.1117 GMCHLA1 0.20261 0.1042 SIGMA(1) 0.64627 0.8627E-02 RHO(1,2) 0.35777E-01 0.9503E-01  t-ratio -7.624 -0.975 -1.116 -1.480 -1.725 -1.912 -5.503 1.927 7.682 -9.261 -4.876 2.456 0.836 -0.415 -0.915 -0.743 1.576 -1.210 -0.053 0.410 -1.151 2.240 -2.791 4.864 0.691 -0.856 -1.264 2.199 4.233 1.944 74.912 0.376  Prob. 0.0000 0.3297 0.2645 0.1390 0.0845 0.0559 0.0000 0.0540 0.0000 0.0000 0.0000 0.0141 0.4030 0.6781 0.3600 0.4576 0.1151 0.2261 0.9577 0.6815 0.2498 0.0251 0.0053 0.0000 0.4893 0.3920 0.2061 0.0279 0.0000 0.0519 0.0000 0.7066  Var. Mean  Var. st. dev.  Economic Valuation Of Critical Habitat Closures, Berman et al.  85  Pacific cod, CPUE Sample Selection Model Two stage least squares regression. Dep. Variable LOGPCOD Observations 2084 Weights ONE Mean of LHS 0.2972981E+00 Std.Dev of LHS 0.3440143E+00 StdDev of resid. 0.3100060E+00 Sum of squares R-squared 0.1875518E+00 Adj. R-squared 0.1756797E+00 F[ 30, 2053] 0.1579768E+02 Prob value 0.3217295E-13 Log-likelihood -0.5007468E+03 Restr.(b=0) Log-l Amemiya Pr. Criter. 0.9753332E-01 Akaike Info.Crit. Standard error corrected for selection..... 0.32875 Correlation of disturbance in regression and Selection Criterion (Rho).............. -0.39356 N(0,1) used for significance levels. Variable Coefficient Std. Error t-ratio Prob. Constant -1.2723 0.4103 -3.101 0.0019 JUN 0.11579 0.7965E-01 1.454 0.1460 JUL -0.41659E-01 0.1154 -0.361 0.7181 AUG -0.58546E-01 0.1618 -0.362 0.7175 SEP 0.93405E-01 0.2067 0.452 0.6514 OCT 0.67453E-01 0.2510 0.269 0.7881 GOA 1.4409 0.3975 3.625 0.0003 TIMELDEP -0.17606E-01 0.1107E-01 -1.590 0.1118 LDEPTH 0.81896 0.1703 4.809 0.0000 L2DEP -0.78711E-01 0.1749E-01 -4.501 0.0000 SLOPE 0.47438E-02 0.2890E-02 1.642 0.1007 SLOPE2 -0.15830E-03 0.6482E-04 -2.442 0.0146 SST 0.43391E-01 0.2125E-01 2.042 0.0411 SST2 -0.10287E-02 0.1575E-02 -0.653 0.5137 SSTSLOPE 0.83700E-05 0.5032E-03 0.017 0.9867 SSH -0.68258E-02 0.3352E-02 -2.036 0.0417 SSHSLOPE 0.12321E-02 0.4771E-03 2.582 0.0098 MWIND 0.19648E-01 0.7717E-01 0.255 0.7990 MCHLA -0.75122E-01 0.2270E-01 -3.309 0.0009 MCHLA1 -0.28885E-01 0.1600E-01 -1.805 0.0711 DWIND 0.27416 0.4677E-01 5.862 0.0000 DCHLA 0.45396E-01 0.3648E-01 1.244 0.2134 DCHLA1 -0.75853E-02 0.3905E-01 -0.194 0.8460 GLDEPTH -0.36074 0.4416E-01 -8.169 0.0000 GTIMELDE 0.21356E-01 0.4631E-02 4.612 0.0000 GSST -0.33675E-01 0.1607E-01 -2.096 0.0361 GSSH 0.25757E-01 0.5804E-02 4.438 0.0000 GMWIND -0.12599 0.1353 -0.931 0.3518 GMCHLA 0.10916 0.4694E-01 2.325 0.0200 GMCHLA1 0.42643E-01 0.4329E-01 0.985 0.3246 IMR7 -0.12938 0.2677E-01 -4.833 0.0000  0.1973010E+03 -0.7327900E+03 0.5103137E+00  Var. Mean  Var. st. dev.  0.0614 0.2692 0.2116 0.1569 0.1934 0.2404 35.5014 4.5578 21.1029 4.3812 106.1247 8.1178 71.7304 20.1822 -3.7809 22.9680 2.0168 1.0076 1.0345 0.0288 0.0509 0.0398 1.1783 8.5557 2.3308 -0.8258 0.4961 0.2574 0.2479 1.1542  0.2402 0.4436 0.4085 0.3638 0.3950 0.4274 7.3817 0.5740 5.4919 9.3259 364.6537 2.4189 37.6320 15.0117 3.8750 16.3217 0.2648 0.4184 0.5422 0.1673 0.2198 0.1956 2.1088 15.7173 4.3117 2.2383 0.8907 0.4944 0.4876 0.4255  Economic Valuation Of Critical Habitat Closures, Berman et al.  86  Pacific cod, CPUE (cont.) Limited Dependent Variable Model - CENSORED Maximum Likelihood Estimates Log-Likelihood.............. -752.97 Threshold values for the model: Lower N(0,1) used for significance levels. Variable Coefficient Std. Error Constant -3.7195 0.5497 JUN 0.73074E-01 0.9327E-01 JUL -0.94603E-01 0.1408 AUG -0.16166 0.2004 SEP -0.51867E-01 0.2574 OCT -0.86989E-01 0.3124 GOA 2.2360 0.4688 TIMELDEP -0.12539E-01 0.1385E-01 LDEPTH 1.9559 0.2269 L2DEP -0.21204 0.2374E-01 SLOPE 0.52226E-02 0.3241E-02 SLOPE2 -0.16501E-03 0.7142E-04 SST 0.40015E-01 0.2453E-01 SST2 -0.66932E-03 0.1811E-02 SSTSLOPE 0.41485E-03 0.5517E-03 SSH -0.51768E-02 0.3694E-02 SSHSLOPE 0.88669E-03 0.5294E-03 MWIND 0.38426E-01 0.8328E-01 MCHLA -0.67076E-01 0.2473E-01 MCHLA1 -0.35456E-01 0.1760E-01 DWIND 0.28878 0.5112E-01 DCHLA 0.36174E-01 0.4144E-01 DCHLA1 -0.19423E-01 0.4524E-01 GLDEPTH -0.49552 0.5405E-01 GTIMELDE 0.21449E-01 0.5321E-02 GSST -0.48657E-01 0.1829E-01 GSSH 0.31160E-01 0.6686E-02 GMWIND -0.16473 0.1553 GMCHLA 0.14357 0.5368E-01 GMCHLA1 0.63966E-01 0.4930E-01 IMR7 -0.11801 0.2947E-01 Sigma 0.33502 0.5614E-02  regression 0.0000  Upper  **********  t-ratio -6.767 0.783 -0.672 -0.807 -0.201 -0.278 4.770 -0.905 8.621 -8.933 1.611 -2.310 1.632 -0.370 0.752 -1.401 1.675 0.461 -2.712 -2.015 5.649 0.873 -0.429 -9.168 4.031 -2.660 4.660 -1.061 2.675 1.297 -4.005 59.678  Prob. 0.0000 0.4334 0.5017 0.4198 0.8403 0.7807 0.0000 0.3654 0.0000 0.0000 0.1071 0.0209 0.1028 0.7117 0.4521 0.1611 0.0940 0.6445 0.0067 0.0439 0.0000 0.3827 0.6677 0.0000 0.0001 0.0078 0.0000 0.2888 0.0075 0.1945 0.0001 0.0000  Var. Mean  Var. st. dev.  0.0614 0.2694 0.2114 0.1572 0.1932 0.2407 35.5039 4.5580 21.1043 4.3922 106.5008 8.1200 71.7742 20.1927 -3.7785 22.9994 2.0166 1.0074 1.0342 0.0288 0.0508 0.0398 1.1795 8.5635 2.3350 -0.8271 0.4965 0.2577 0.2481 1.9112  0.2400 0.4438 0.4084 0.3641 0.3949 0.4276 7.3800 0.5738 5.4895 9.3408 365.0076 2.4204 37.6846 15.0243 3.8752 16.3473 0.2648 0.4183 0.5422 0.1672 0.2197 0.1955 2.1094 15.7202 4.3166 2.2387 0.8907 0.4945 0.4877 0.4671  Economic Valuation Of Critical Habitat Closures, Berman et al.  87  Pacific cod, CPUE (cont.) ML Estimates of Selection Model Maximum Likelihood Estimates Log-Likelihood.............. -5068.9 LHS is CENSORED. Tobit Model fit by MLE. FIRST 30 estimates are probit equation. N(0,1) used for significance levels. Variable Coefficient Std. Error Constant -4.2442 0.6238 JUN 0.60673E-01 0.9264E-01 JUL -0.10202 0.1355 AUG -0.17891 0.1959 SEP -0.81157E-01 0.2507 OCT -0.11978 0.3072 GOA 2.2362 0.4178 TIMELDEP -0.97058E-02 0.1348E-01 LDEPTH 2.1088 0.2630 L2DEP -0.22970 0.2727E-01 SLOPE 0.61017E-02 0.3329E-02 SLOPE2 -0.17241E-03 0.7232E-04 SST 0.44229E-01 0.2528E-01 SST2 -0.79041E-03 0.1924E-02 SSTSLOPE 0.46403E-03 0.6213E-03 SSH -0.52626E-02 0.3588E-02 SSHSLOPE 0.10845E-02 0.5087E-03 MWIND 0.22658E-01 0.8935E-01 MCHLA -0.63126E-01 0.2815E-01 MCHLA1 -0.30799E-01 0.2229E-01 DWIND 0.27695 0.3481E-01 DCHLA 0.30429E-01 0.5056E-01 DCHLA1 -0.31148E-01 0.4778E-01 GLDEPTH -0.49579 0.5434E-01 GTIMELDE 0.20498E-01 0.5050E-02 GSST -0.49139E-01 0.1878E-01 GSSH 0.31766E-01 0.5106E-02 GMWIND -0.15793 0.1317 GMCHLA 0.14375 0.4085E-01 GMCHLA1 0.61543E-01 0.4136E-01 SIGMA(1) 0.34485 0.7897E-02 RHO(1,2) -0.27904 0.9210E-01  t-ratio -6.804 0.655 -0.753 -0.913 -0.324 -0.390 5.352 -0.720 8.018 -8.424 1.833 -2.384 1.750 -0.411 0.747 -1.467 2.132 0.254 -2.243 -1.382 7.956 0.602 -0.652 -9.123 4.059 -2.616 6.221 -1.199 3.519 1.488 43.669 -3.030  Prob. 0.0000 0.5125 0.4516 0.3611 0.7461 0.6966 0.0000 0.4717 0.0000 0.0000 0.0668 0.0171 0.0802 0.6812 0.4551 0.1425 0.0330 0.7998 0.0249 0.1670 0.0000 0.5473 0.5144 0.0000 0.0000 0.0089 0.0000 0.2306 0.0004 0.1367 0.0000 0.0024  Var. Mean  Var. st. dev.  88  Economic Valuation Of Critical Habitat Closures, Berman et al.  Pacific cod, standard CPUE Sample Selection Model Two stage least squares regression. Dep. Variable LOGPCODS Observations 2084 Weights ONE Mean of LHS 0.4752573E+00 Std.Dev of LHS 0.5026082E+00 StdDev of resid. 0.4405847E+00 Sum of squares R-squared 0.2312093E+00 Adj. R-squared 0.2199752E+00 F[ 30, 2053] 0.2058093E+02 Prob value Log-likelihood -0.1233296E+04 Restr.(b=0) Log-l Amemiya Pr. Criter. 0.1970024E+00 Akaike Info.Crit. Standard error corrected for selection..... 0.47276 Correlation of disturbance in regression and Selection Criterion (Rho).............. -0.42883 N(0,1) used for significance levels. Variable Coefficient Std. Error t-ratio Prob. Constant -3.0312 0.5839 -5.191 0.0000 JUN 0.10253E-01 0.1133 0.090 0.9279 JUL -0.34272 0.1640 -2.089 0.0367 AUG -0.49949 0.2300 -2.172 0.0299 SEP -0.43265 0.2937 -1.473 0.1408 OCT -0.63212 0.3566 -1.773 0.0763 GOA 1.5999 0.5649 2.832 0.0046 TIMELDEP 0.74061E-02 0.1573E-01 0.471 0.6378 LDEPTH 1.4534 0.2422 6.001 0.0000 L2DEP -0.15923 0.2490E-01 -6.395 0.0000 SLOPE -0.47634E-02 0.4127E-02 -1.154 0.2484 SLOPE2 -0.87827E-04 0.9267E-04 -0.948 0.3433 SST 0.98077E-01 0.3022E-01 3.246 0.0012 SST2 -0.36883E-02 0.2240E-02 -1.647 0.0996 SSTSLOPE 0.35150E-03 0.7179E-03 0.490 0.6244 SSH 0.63242E-04 0.4775E-02 0.013 0.9894 SSHSLOPE 0.15362E-02 0.6806E-03 2.257 0.0240 MWIND 0.10685 0.1100 0.972 0.3313 MCHLA -0.12999 0.3234E-01 -4.020 0.0001 MCHLA1 -0.36242E-01 0.2283E-01 -1.588 0.1123 DWIND 0.38428 0.6654E-01 5.775 0.0000 DCHLA 0.12941 0.5197E-01 2.490 0.0128 DCHLA1 -0.27154E-01 0.5566E-01 -0.488 0.6257 GLDEPTH -0.44470 0.6278E-01 -7.084 0.0000 GTIMELDE 0.25827E-01 0.6586E-02 3.921 0.0001 GSST -0.35985E-01 0.2283E-01 -1.576 0.1150 GSSH 0.35849E-01 0.8251E-02 4.345 0.0000 GMWIND -0.37602E-01 0.1924 -0.195 0.8450 GMCHLA 0.10319 0.6673E-01 1.546 0.1220 GMCHLA1 0.71789E-01 0.6151E-01 1.167 0.2432 IMR7 -0.20273 0.3795E-01 -5.342 0.0000  0.3985179E+03 0.3217295E-13 -0.1522892E+04 0.1213336E+01  Var. Mean  Var. st. dev.  0.0614 0.2692 0.2116 0.1569 0.1934 0.2404 35.5014 4.5578 21.1029 4.3812 106.1247 8.1178 71.7304 20.1822 -3.7809 22.9680 2.0168 1.0076 1.0345 0.0288 0.0509 0.0398 1.1783 8.5557 2.3308 -0.8258 0.4961 0.2574 0.2479 1.1540  0.2402 0.4436 0.4085 0.3638 0.3950 0.4274 7.3817 0.5740 5.4919 9.3259 364.6537 2.4189 37.6320 15.0117 3.8750 16.3217 0.2648 0.4184 0.5422 0.1673 0.2198 0.1956 2.1088 15.7173 4.3117 2.2383 0.8907 0.4944 0.4876 0.4264  Economic Valuation Of Critical Habitat Closures, Berman et al.  89  Pacific cod, standard CPUE (cont.) Limited Dependent Variable Model - CENSORED Maximum Likelihood Estimates Log-Likelihood.............. -1400.0 Threshold values for the model: Lower N(0,1) used for significance levels. Variable Coefficient Std. Error Constant -6.3560 0.7936 JUN -0.62224E-01 0.1325 JUL -0.46524 0.1995 AUG -0.70861 0.2837 SEP -0.71877 0.3645 OCT -0.95623 0.4424 GOA 2.5754 0.6638 TIMELDEP 0.19013E-01 0.1958E-01 LDEPTH 3.0260 0.3295 L2DEP -0.34505 0.3424E-01 SLOPE -0.64795E-02 0.4597E-02 SLOPE2 -0.75276E-04 0.1015E-03 SST 0.86105E-01 0.3538E-01 SST2 -0.28320E-02 0.2598E-02 SSTSLOPE 0.90923E-03 0.7848E-03 SSH 0.24922E-02 0.5253E-02 SSHSLOPE 0.99416E-03 0.7513E-03 MWIND 0.14132 0.1184 MCHLA -0.12416 0.3519E-01 MCHLA1 -0.53522E-01 0.2503E-01 DWIND 0.40964 0.7310E-01 DCHLA 0.13303 0.5882E-01 DCHLA1 -0.29947E-01 0.6410E-01 GLDEPTH -0.61108 0.7630E-01 GTIMELDE 0.26484E-01 0.7525E-02 GSST -0.55655E-01 0.2602E-01 GSSH 0.42989E-01 0.9458E-02 GMWIND -0.78227E-01 0.2201 GMCHLA 0.13823 0.7603E-01 GMCHLA1 0.10721 0.6978E-01 IMR7 -0.22882 0.4230E-01 Sigma 0.47643 0.7996E-02  regression 0.0000  Upper  **********  t-ratio -8.009 -0.470 -2.332 -2.497 -1.972 -2.161 3.880 0.971 9.184 -10.076 -1.409 -0.742 2.434 -1.090 1.159 0.474 1.323 1.194 -3.528 -2.138 5.604 2.262 -0.467 -8.009 3.520 -2.139 4.545 -0.355 1.818 1.537 -5.410 59.582  Prob. 0.0000 0.6387 0.0197 0.0125 0.0486 0.0307 0.0001 0.3314 0.0000 0.0000 0.1587 0.4581 0.0149 0.2757 0.2466 0.6352 0.1858 0.2326 0.0004 0.0325 0.0000 0.0237 0.6403 0.0000 0.0004 0.0325 0.0000 0.7223 0.0690 0.1244 0.0000 0.0000  Var. Mean  Var. st. dev.  0.0614 0.2694 0.2114 0.1572 0.1932 0.2407 35.5039 4.5580 21.1043 4.3922 106.5008 8.1200 71.7742 20.1927 -3.7785 22.9994 2.0166 1.0074 1.0342 0.0288 0.0508 0.0398 1.1795 8.5635 2.3350 -0.8271 0.4965 0.2577 0.2481 1.9112  0.2400 0.4438 0.4084 0.3641 0.3949 0.4276 7.3800 0.5738 5.4895 9.3408 365.0076 2.4204 37.6846 15.0243 3.8752 16.3473 0.2648 0.4183 0.5422 0.1672 0.2197 0.1955 2.1094 15.7202 4.3166 2.2387 0.8907 0.4945 0.4877 0.4671  Economic Valuation Of Critical Habitat Closures, Berman et al.  90  Pacific cod, standard CPUE (cont.) ML Estimates of Selection Model Maximum Likelihood Estimates Log-Likelihood.............. -5719.8 LHS is CENSORED. Tobit Model fit by MLE. FIRST 30 estimates are probit equation. N(0,1) used for significance levels. Variable Coefficient Std. Error Constant -7.4990 0.8471 JUN -0.10092 0.1376 JUL -0.48861 0.1929 AUG -0.75684 0.2736 SEP -0.79366 0.3505 OCT -1.0382 0.4283 GOA 2.5939 0.5660 TIMELDEP 0.25656E-01 0.1893E-01 LDEPTH 3.3667 0.3535 L2DEP -0.38555 0.3659E-01 SLOPE -0.41864E-02 0.4831E-02 SLOPE2 -0.92706E-04 0.1096E-03 SST 0.96412E-01 0.3881E-01 SST2 -0.31448E-02 0.2783E-02 SSTSLOPE 0.10493E-02 0.8380E-03 SSH 0.23009E-02 0.5296E-02 SSHSLOPE 0.12845E-02 0.7238E-03 MWIND 0.10022 0.1332 MCHLA -0.10925 0.4291E-01 MCHLA1 -0.42077E-01 0.3249E-01 DWIND 0.37687 0.5157E-01 DCHLA 0.11479 0.6983E-01 DCHLA1 -0.53688E-01 0.6483E-01 GLDEPTH -0.61036 0.7158E-01 GTIMELDE 0.24273E-01 0.7297E-02 GSST -0.57593E-01 0.2552E-01 GSSH 0.44010E-01 0.7312E-02 GMWIND -0.64972E-01 0.1895 GMCHLA 0.13433 0.5970E-01 GMCHLA1 0.10391 0.5838E-01 SIGMA(1) 0.49707 0.1360E-01 RHO(1,2) -0.32736 0.9276E-01  t-ratio -8.852 -0.733 -2.532 -2.766 -2.264 -2.424 4.583 1.355 9.525 -10.538 -0.867 -0.845 2.484 -1.130 1.252 0.434 1.775 0.753 -2.546 -1.295 7.309 1.644 -0.828 -8.527 3.327 -2.257 6.019 -0.343 2.250 1.780 36.545 -3.529  Prob. 0.0000 0.4634 0.0113 0.0057 0.0235 0.0154 0.0000 0.1754 0.0000 0.0000 0.3861 0.3978 0.0130 0.2585 0.2105 0.6640 0.0760 0.4518 0.0109 0.1952 0.0000 0.1002 0.4076 0.0000 0.0009 0.0240 0.0000 0.7318 0.0244 0.0751 0.0000 0.0004  Var. Mean  Var. st. dev.  Economic Valuation Of Critical Habitat Closures, Berman et al.  91  Atka mackerel, standard CPUE Sample Selection Model Two stage least squares regression. Dep. Variable LOGATKAS Observations 2084 Weights ONE Mean of LHS 0.1138042E+00 Std.Dev of LHS 0.5121301E+00 StdDev of resid. 0.3874007E+00 Sum of squares R-squared 0.4275091E+00 Adj. R-squared 0.4191435E+00 F[ 30, 2053] 0.5110278E+02 Prob value 0.3217295E-13 Log-likelihood -0.9652031E+03 Restr.(b=0) Log-l Amemiya Pr. Criter. 0.1523118E+00 Akaike Info.Crit. Standard error corrected for selection..... 0.38994 Correlation of disturbance in regression and Selection Criterion (Rho).............. -0.13468 N(0,1) used for significance levels. Variable Coefficient Std. Error t-ratio Prob. Constant -3.4712 0.5074 -6.841 0.0000 JUN -0.67829 0.9920E-01 -6.838 0.0000 JUL -1.1060 0.1442 -7.673 0.0000 AUG -1.2967 0.2023 -6.411 0.0000 SEP -1.5553 0.2586 -6.015 0.0000 OCT -1.9524 0.3140 -6.219 0.0000 GOA -0.97079 0.4965 -1.955 0.0506 TIMELDEP 0.79617E-01 0.1385E-01 5.747 0.0000 LDEPTH 1.5793 0.2108 7.492 0.0000 L2DEP -0.24881 0.2161E-01 -11.515 0.0000 SLOPE 0.60933E-01 0.3525E-02 17.284 0.0000 SLOPE2 -0.81759E-03 0.7889E-04 -10.364 0.0000 SST 0.71823E-01 0.2647E-01 2.713 0.0067 SST2 -0.70958E-02 0.1965E-02 -3.611 0.0003 SSTSLOPE -0.35416E-03 0.6188E-03 -0.572 0.5671 SSH 0.10969E-01 0.4147E-02 2.645 0.0082 SSHSLOPE 0.13210E-02 0.5863E-03 2.253 0.0242 MWIND -0.13900E-01 0.9491E-01 -0.146 0.8836 MCHLA -0.95928E-01 0.2804E-01 -3.421 0.0006 MCHLA1 -0.77336E-02 0.1969E-01 -0.393 0.6945 DWIND 0.31106 0.5794E-01 5.368 0.0000 DCHLA 0.87823E-01 0.4508E-01 1.948 0.0514 DCHLA1 0.26594 0.4818E-01 5.520 0.0000 GLDEPTH 0.25282E-01 0.5511E-01 0.459 0.6464 GTIMELDE -0.99106E-02 0.5767E-02 -1.718 0.0857 GSST 0.10888 0.2009E-01 5.419 0.0000 GSSH -0.16702E-01 0.7246E-02 -2.305 0.0212 GMWIND -0.64004E-01 0.1689 -0.379 0.7047 GMCHLA 0.47865E-01 0.5866E-01 0.816 0.4145 GMCHLA1 0.81060E-01 0.5412E-01 1.498 0.1342 IMR7 -0.52518E-01 0.3285E-01 -1.599 0.1099  0.3081128E+03 -0.1562004E+04 0.9560491E+00  Var. Mean  Var. st. dev.  0.0614 0.2692 0.2116 0.1569 0.1934 0.2404 35.5014 4.5578 21.1029 4.3812 106.1247 8.1178 71.7304 20.1822 -3.7809 22.9680 2.0168 1.0076 1.0345 0.0288 0.0509 0.0398 1.1783 8.5557 2.3308 -0.8258 0.4961 0.2574 0.2479 1.1538  0.2402 0.4436 0.4085 0.3638 0.3950 0.4274 7.3817 0.5740 5.4919 9.3259 364.6537 2.4189 37.6320 15.0117 3.8750 16.3217 0.2648 0.4184 0.5422 0.1673 0.2198 0.1956 2.1088 15.7173 4.3117 2.2383 0.8907 0.4944 0.4876 0.4263  Economic Valuation Of Critical Habitat Closures, Berman et al.  92  Atka mackerel, standard CPUE (cont.) Limited Dependent Variable Model - CENSORED Maximum Likelihood Estimates Log-Likelihood.............. -499.01 Threshold values for the model: Lower N(0,1) used for significance levels. Variable Coefficient Std. Error Constant -71.966 9.669 JUN -0.53299 0.7549 JUL -0.26201 1.205 AUG 0.70599 1.787 SEP 1.4722 2.293 OCT 1.4626 2.851 GOA 3.3275 3.532 TIMELDEP -0.57410E-01 0.1199 LDEPTH 28.726 4.040 L2DEP -2.8680 0.3611 SLOPE 0.16240 0.1591E-01 SLOPE2 -0.23148E-02 0.3170E-03 SST 0.35657 0.1572 SST2 -0.46840E-01 0.1205E-01 SSTSLOPE 0.35578E-02 0.4259E-02 SSH 0.48134E-01 0.2299E-01 SSHSLOPE 0.11133E-02 0.3546E-02 MWIND 0.45062 0.5215 MCHLA -0.34119 0.2358 MCHLA1 -0.33524 0.1792 DWIND 0.40500 0.2792 DCHLA 0.33889 0.2470 DCHLA1 0.74358 0.2210 GLDEPTH -1.2782 0.4338 GTIMELDE -0.35227E-01 0.3543E-01 GSST 0.53089 0.1346 GSSH -0.97291E-01 0.5316E-01 GMWIND 0.19559 1.123 GMCHLA -0.98987 0.4409 GMCHLA1 -0.17773 0.4437 IMR7 -0.23128E-01 0.1741 Sigma 1.1871 0.6330E-01  regression 0.0000  Upper  **********  t-ratio -7.443 -0.706 -0.217 0.395 0.642 0.513 0.942 -0.479 7.110 -7.943 10.207 -7.303 2.268 -3.887 0.835 2.094 0.314 0.864 -1.447 -1.870 1.451 1.372 3.365 -2.947 -0.994 3.943 -1.830 0.174 -2.245 -0.401 -0.133 18.752  Prob. 0.0000 0.4801 0.8279 0.6928 0.5209 0.6079 0.3461 0.6320 0.0000 0.0000 0.0000 0.0000 0.0233 0.0001 0.4035 0.0363 0.7535 0.3875 0.1478 0.0614 0.1469 0.1701 0.0008 0.0032 0.3201 0.0001 0.0672 0.8617 0.0248 0.6887 0.8943 0.0000  Var. Mean  Var. st. dev.  0.0633 0.2678 0.2117 0.1560 0.1946 0.2436 35.4982 4.5562 21.0860 4.3530 105.2351 8.1293 71.9170 20.1658 -3.7812 22.9580 2.0168 1.0088 1.0362 0.0285 0.0504 0.0395 1.1915 8.6620 2.3647 -0.8373 0.5022 0.2617 0.2523 1.9113  0.2435 0.4429 0.4086 0.3630 0.3960 0.4293 7.3771 0.5722 5.4745 9.2912 363.2146 2.4187 37.6799 14.9954 3.8648 16.2960 0.2650 0.4175 0.5414 0.1666 0.2189 0.1948 2.1141 15.7768 4.3352 2.2455 0.8939 0.4980 0.4920 0.4659  Economic Valuation Of Critical Habitat Closures, Berman et al.  93  Atka mackerel, standard CPUE (cont.) ML Estimates of Selection Model Maximum Likelihood Estimates Log-Likelihood.............. -4808.5 LHS is CENSORED. Tobit Model fit by MLE. FIRST 30 estimates are probit equation. N(0,1) used for significance levels. Variable Coefficient Std. Error Constant -54.075 8.874 JUN 0.35337 0.8760 JUL 0.11495 1.508 AUG 1.2717 2.289 SEP 1.7955 2.963 OCT 2.0186 3.662 GOA 1.0775 4.573 TIMELDEP -0.10824 0.1534 LDEPTH 21.902 3.711 L2DEP -2.1076 0.3086 SLOPE 0.13907 0.1884E-01 SLOPE2 -0.23223E-02 0.4010E-03 SST 0.15489 0.2127 SST2 -0.31409E-01 0.1520E-01 SSTSLOPE -0.21312E-02 0.4937E-02 SSH 0.39028E-01 0.2478E-01 SSHSLOPE 0.28872E-02 0.4043E-02 MWIND 1.1562 0.5865 MCHLA -0.67795 0.2351 MCHLA1 -0.49768 0.1921 DWIND 0.91228 0.3394 DCHLA 0.73018 0.2850 DCHLA1 0.87158 0.2442 GLDEPTH -1.1428 0.5303 GTIMELDE -0.13840E-01 0.4993E-01 GSST 0.48191 0.1817 GSSH -0.61423E-01 0.7320E-01 GMWIND 0.64325 1.615 GMCHLA -0.72321 0.5613 GMCHLA1 -0.85615E-01 0.5399 SIGMA(1) 1.6226 0.1807 RHO(1,2) -0.82007 0.5919E-01  t-ratio -6.094 0.403 0.076 0.556 0.606 0.551 0.236 -0.706 5.901 -6.830 7.380 -5.792 0.728 -2.066 -0.432 1.575 0.714 1.972 -2.883 -2.591 2.688 2.562 3.570 -2.155 -0.277 2.652 -0.839 0.398 -1.289 -0.159 8.979 -13.855  Prob. 0.0000 0.6867 0.9392 0.5785 0.5445 0.5814 0.8137 0.4804 0.0000 0.0000 0.0000 0.0000 0.4665 0.0389 0.6660 0.1153 0.4752 0.0487 0.0039 0.0096 0.0072 0.0104 0.0004 0.0312 0.7816 0.0080 0.4014 0.6905 0.1976 0.8740 0.0000 0.0000  Var. Mean  Var. st. dev.  94  Economic Valuation Of Critical Habitat Closures, Berman et al.  Black cod, standard CPUE Sample Selection Model Two stage least squares regression. Dep. Variable LOGBCODS Observations 2084 Weights ONE Mean of LHS 0.7045839E-01 Std.Dev of LHS 0.2518961E+00 StdDev of resid. 0.2016822E+00 Sum of squares 0.8350725E+02 R-squared 0.3586414E+00 Adj. R-squared 0.3492694E+00 F[ 30, 2053] 0.3826724E+02 Prob value 0.3217295E-13 Log-likelihood 0.3951616E+03 Restr.(b=0) Log-l -0.8327635E+02 Amemiya Pr. Criter. 0.4128078E-01 Akaike Info.Crit. -0.3494833E+00 Standard error corrected for selection..... 0.21504 Correlation of disturbance in regression and Selection Criterion (Rho).............. -0.41018 N(0,1) used for significance levels. Variable Coefficient Std. Error t-ratio Prob. Var. Mean Var. st. dev. Constant 1.8098 0.2727 6.635 0.0000 JUN 0.14077 0.5192E-01 2.711 0.0067 0.0614 0.2402 JUL 0.16869 0.7505E-01 2.248 0.0246 0.2692 0.4436 AUG 0.19537 0.1052 1.857 0.0633 0.2116 0.4085 SEP 0.18826 0.1344 1.400 0.1614 0.1569 0.3638 OCT 0.16351 0.1632 1.002 0.3164 0.1934 0.3950 GOA -1.0550 0.2586 -4.079 0.0000 0.2404 0.4274 TIMELDEP -0.49002E-02 0.7198E-02 -0.681 0.4960 35.5014 7.3817 LDEPTH -0.90681 0.1134 -7.995 0.0000 4.5578 0.5740 L2DEP 0.11537 0.1170E-01 9.863 0.0000 21.1029 5.4919 SLOPE -0.69613E-02 0.1903E-02 -3.659 0.0003 4.3812 9.3259 SLOPE2 0.45571E-04 0.4235E-04 1.076 0.2819 106.1247 364.6537 SST -0.29274E-01 0.1384E-01 -2.115 0.0344 8.1178 2.4189 SST2 0.86234E-03 0.1025E-02 0.841 0.4003 71.7304 37.6320 SSTSLOPE 0.52476E-04 0.3278E-03 0.160 0.8728 20.1822 15.0117 SSH -0.42687E-04 0.2183E-02 -0.020 0.9844 -3.7809 3.8750 SSHSLOPE 0.58802E-03 0.3113E-03 1.889 0.0589 22.9680 16.3217 MWIND 0.11473 0.5041E-01 2.276 0.0229 2.0168 0.2648 MCHLA -0.15123E-02 0.1481E-01 -0.102 0.9187 1.0076 0.4184 MCHLA1 -0.18364E-01 0.1042E-01 -1.762 0.0781 1.0345 0.5422 DWIND 0.53786E-01 0.3039E-01 1.770 0.0767 0.0288 0.1673 DCHLA 0.53918E-01 0.2379E-01 2.266 0.0234 0.0509 0.2198 DCHLA1 0.46058E-01 0.2544E-01 1.810 0.0703 0.0398 0.1956 GLDEPTH 0.20449 0.2873E-01 7.118 0.0000 1.1783 2.1088 GTIMELDE 0.11063E-02 0.3020E-02 0.366 0.7141 8.5557 15.7173 GSST 0.34821E-01 0.1044E-01 3.334 0.0009 2.3308 4.3117 GSSH -0.49733E-02 0.3780E-02 -1.316 0.1883 -0.8258 2.2383 GMWIND 0.22938E-02 0.8804E-01 0.026 0.9792 0.4961 0.8907 GMCHLA -0.74513E-01 0.3051E-01 -2.442 0.0146 0.2574 0.4944 GMCHLA1 -0.63518E-01 0.2817E-01 -2.255 0.0241 0.2479 0.4876 IMR7 -0.88207E-01 0.1802E-01 -4.896 0.0000 1.1568 0.4254  Economic Valuation Of Critical Habitat Closures, Berman et al.  95  Black cod, standard CPUE (cont.) Limited Dependent Variable Model - CENSORED Maximum Likelihood Estimates Log-Likelihood.............. -475.33 Threshold values for the model: Lower N(0,1) used for significance levels. Variable Coefficient Std. Error Constant 1.2131 2.136 JUN 0.54405 0.3012 JUL 0.45107 0.3643 AUG 0.42911 0.5001 SEP 0.58552 0.6314 OCT 0.47957 0.7677 GOA -1.4643 0.8430 TIMELDEP 0.20330E-01 0.3037E-01 LDEPTH -1.0072 0.8306 L2DEP 0.18929 0.7475E-01 SLOPE -0.94604E-02 0.7722E-02 SLOPE2 -0.43875E-03 0.2070E-03 SST -0.13363 0.7330E-01 SST2 0.58258E-02 0.4761E-02 SSTSLOPE -0.19264E-02 0.1486E-02 SSH -0.21699E-01 0.1219E-01 SSHSLOPE 0.26229E-02 0.1163E-02 MWIND -0.41253 0.1958 MCHLA -0.24595 0.1010 MCHLA1 -0.10473 0.6237E-01 DWIND 0.37205 0.1130 DCHLA 0.28535 0.1020 DCHLA1 0.14997 0.9825E-01 GLDEPTH 0.32492 0.1238 GTIMELDE -0.43333E-01 0.1475E-01 GSST 0.52365E-01 0.3993E-01 GSSH 0.19796E-01 0.1484E-01 GMWIND 0.95612 0.3191 GMCHLA 0.15454 0.1236 GMCHLA1 -0.13879E-01 0.9517E-01 IMR7 -0.59326 0.9035E-01 Sigma 0.48232 0.1793E-01  regression 0.0000  Upper  **********  t-ratio 0.568 1.806 1.238 0.858 0.927 0.625 -1.737 0.669 -1.213 2.532 -1.225 -2.120 -1.823 1.224 -1.297 -1.779 2.255 -2.107 -2.436 -1.679 3.293 2.797 1.527 2.625 -2.937 1.311 1.334 2.996 1.250 -0.146 -6.566 26.899  Prob. 0.5700 0.0709 0.2157 0.3908 0.3538 0.5322 0.0824 0.5033 0.2253 0.0113 0.2205 0.0340 0.0683 0.2211 0.1948 0.0752 0.0241 0.0351 0.0149 0.0931 0.0010 0.0052 0.1269 0.0087 0.0033 0.1898 0.1824 0.0027 0.2112 0.8841 0.0000 0.0000  Var. Mean  Var. st. dev.  0.0632 0.2679 0.2114 0.1563 0.1943 0.2437 35.4959 4.5567 21.0905 4.3824 106.1783 8.1266 71.8882 20.1915 -3.7830 23.0019 2.0167 1.0086 1.0359 0.0290 0.0504 0.0399 1.1919 8.6648 2.3660 -0.8388 0.5024 0.2619 0.2526 1.9115  0.2433 0.4430 0.4084 0.3632 0.3958 0.4294 7.3762 0.5722 5.4739 9.3282 364.3158 2.4220 37.6890 15.0350 3.8678 16.3287 0.2648 0.4172 0.5412 0.1678 0.2187 0.1958 2.1142 15.7758 4.3357 2.2465 0.8940 0.4981 0.4922 0.4680  Economic Valuation Of Critical Habitat Closures, Berman et al.  96  Black cod, standard CPUE (cont.) ML Estimates of Selection Model Maximum Likelihood Estimates Log-Likelihood.............. -4800.0 LHS is CENSORED. Tobit Model fit by MLE. FIRST 30 estimates are probit equation. N(0,1) used for significance levels. Variable Coefficient Std. Error Constant -1.8232 2.353 JUN 0.55945 0.4032 JUL 0.48645 0.5183 AUG 0.48453 0.7006 SEP 0.52862 0.8851 OCT 0.35517 1.074 GOA -1.3881 1.064 TIMELDEP 0.35108E-01 0.4106E-01 LDEPTH -0.19972 0.9987 L2DEP 0.93209E-01 0.9095E-01 SLOPE -0.17968E-02 0.8992E-02 SLOPE2 -0.48634E-03 0.2269E-03 SST -0.11236 0.1066 SST2 0.43708E-02 0.6897E-02 SSTSLOPE -0.15135E-02 0.1960E-02 SSH -0.27004E-01 0.1562E-01 SSHSLOPE 0.32944E-02 0.1333E-02 MWIND -0.41676 0.2745 MCHLA -0.20416 0.1667 MCHLA1 -0.60272E-01 0.9581E-01 DWIND 0.29563 0.1353 DCHLA 0.23332 0.1148 DCHLA1 0.89461E-01 0.1318 GLDEPTH 0.35292 0.1715 GTIMELDE -0.50061E-01 0.2251E-01 GSST 0.56143E-01 0.6438E-01 GSSH 0.26269E-01 0.1766E-01 GMWIND 0.94383 0.4029 GMCHLA 0.13359 0.1906 GMCHLA1 -0.42591E-01 0.1296 SIGMA(1) 0.60758 0.6781E-01 RHO(1,2) -0.65815 0.1227  t-ratio -0.775 1.387 0.938 0.692 0.597 0.331 -1.305 0.855 -0.200 1.025 -0.200 -2.143 -1.054 0.634 -0.772 -1.729 2.472 -1.518 -1.225 -0.629 2.185 2.032 0.679 2.057 -2.224 0.872 1.488 2.343 0.701 -0.329 8.961 -5.366  Prob. 0.4384 0.1653 0.3480 0.4892 0.5504 0.7409 0.1919 0.3925 0.8415 0.3054 0.8416 0.0321 0.2918 0.5262 0.4401 0.0839 0.0134 0.1289 0.2207 0.5293 0.0289 0.0422 0.4972 0.0397 0.0261 0.3831 0.1368 0.0191 0.4833 0.7425 0.0000 0.0000  Var. Mean  Var. st. dev.  Economic Valuation Of Critical Habitat Closures, Berman et al.  Rockfish, standard CPUE Sample Selection Model Two stage least squares regression. Dep. Variable LOGROCKS Observations 2084 Weights ONE Mean of LHS 0.3189948E+00 Std.Dev of LHS 0.8042483E+00 StdDev of resid. 0.5704857E+00 Sum of squares 0.6681570E+03 R-squared 0.4965948E+00 Adj. R-squared 0.4892387E+00 F[ 30, 2053] 0.6750753E+02 Prob value 0.3217295E-13 Log-likelihood -0.1771771E+04 Restr.(b=0) Log-l -0.2502574E+04 Amemiya Pr. Criter. 0.3302952E+00 Akaike Info.Crit. 0.1730106E+01 Standard error corrected for selection..... 0.57731 Correlation of disturbance in regression and Selection Criterion (Rho).............. 0.18122 N(0,1) used for significance levels. Variable Coefficient Std. Error t-ratio Prob. Var. Mean Var. st. dev. Constant -3.1113 0.7469 -4.165 0.0000 JUN -0.50678E-01 0.1460 -0.347 0.7285 0.0614 0.2402 JUL 0.18668 0.2123 0.879 0.3792 0.2692 0.4436 AUG -0.19276 0.2978 -0.647 0.5175 0.2116 0.4085 SEP -0.74679E-01 0.3807 -0.196 0.8445 0.1569 0.3638 OCT 0.28892 0.4623 0.625 0.5320 0.1934 0.3950 GOA -1.9658 0.7314 -2.688 0.0072 0.2404 0.4274 TIMELDEP 0.42586E-02 0.2040E-01 0.209 0.8346 35.5014 7.3817 LDEPTH 1.5964 0.3102 5.147 0.0000 4.5578 0.5740 L2DEP -0.16010 0.3176E-01 -5.041 0.0000 21.1029 5.4919 SLOPE 0.50755E-01 0.5203E-02 9.754 0.0000 4.3812 9.3259 SLOPE2 -0.45105E-03 0.1165E-03 -3.872 0.0001 106.1247 364.6537 SST -0.17723 0.3899E-01 -4.546 0.0000 8.1178 2.4189 SST2 0.14955E-01 0.2894E-02 5.167 0.0000 71.7304 37.6320 SSTSLOPE 0.13599E-02 0.9128E-03 1.490 0.1363 20.1822 15.0117 SSH 0.23040E-01 0.6114E-02 3.769 0.0002 -3.7809 3.8750 SSHSLOPE 0.13567E-02 0.8649E-03 1.569 0.1167 22.9680 16.3217 MWIND -0.26801 0.1398 -1.917 0.0552 2.0168 0.2648 MCHLA 0.23437E-01 0.4124E-01 0.568 0.5698 1.0076 0.4184 MCHLA1 -0.10672 0.2902E-01 -3.678 0.0002 1.0345 0.5422 DWIND 0.64795E-01 0.8539E-01 0.759 0.4480 0.0288 0.1673 DCHLA 0.24483 0.6637E-01 3.689 0.0002 0.0509 0.2198 DCHLA1 0.32186 0.7106E-01 4.529 0.0000 0.0398 0.1956 GLDEPTH 0.35931 0.8116E-01 4.427 0.0000 1.1783 2.1088 GTIMELDE -0.31584E-01 0.8495E-02 -3.718 0.0002 8.5557 15.7173 GSST 0.89732E-01 0.2959E-01 3.033 0.0024 2.3308 4.3117 GSSH -0.39775E-01 0.1067E-01 -3.728 0.0002 -0.8258 2.2383 GMWIND 0.18776 0.2487 0.755 0.4503 0.4961 0.8907 GMCHLA -0.12895 0.8632E-01 -1.494 0.1352 0.2574 0.4944 GMCHLA1 0.19343 0.7970E-01 2.427 0.0152 0.2479 0.4876 IMR7 0.10462 0.4816E-01 2.172 0.0298 1.1547 0.4262  97  Economic Valuation Of Critical Habitat Closures, Berman et al.  98  Rockfish, standard CPUE (cont.) Limited Dependent Variable Model - CENSORED Maximum Likelihood Estimates Log-Likelihood.............. -1058.7 Threshold values for the model: Lower N(0,1) used for significance levels. Variable Coefficient Std. Error Constant -59.038 4.225 JUN -0.15847E-01 0.4224 JUL 1.2045 0.6004 AUG 1.1343 0.8765 SEP 2.3211 1.119 OCT 3.1534 1.383 GOA 5.7371 1.695 TIMELDEP -0.84149E-01 0.5672E-01 LDEPTH 21.942 1.623 L2DEP -1.9408 0.1388 SLOPE 0.89567E-01 0.9907E-02 SLOPE2 -0.97000E-03 0.2125E-03 SST -0.18383 0.9804E-01 SST2 0.11912E-01 0.7003E-02 SSTSLOPE 0.49697E-02 0.2508E-02 SSH 0.32366E-01 0.1621E-01 SSHSLOPE 0.24707E-02 0.2035E-02 MWIND -0.46247 0.3564 MCHLA 0.45035 0.1383 MCHLA1 -0.29367 0.9248E-01 DWIND 0.35857E-01 0.1803 DCHLA 0.12472 0.1533 DCHLA1 0.58706 0.1424 GLDEPTH -0.96351 0.1986 GTIMELDE -0.56791E-01 0.1835E-01 GSST 0.18872 0.6882E-01 GSSH -0.69470E-01 0.2373E-01 GMWIND -0.18797E-01 0.5551 GMCHLA -0.55850 0.1975 GMCHLA1 0.30757 0.1641 IMR7 0.33952 0.1143 Sigma 0.98623 0.2846E-01  regression 0.0000  Upper  **********  t-ratio -13.975 -0.038 2.006 1.294 2.074 2.281 3.385 -1.484 13.521 -13.979 9.040 -4.565 -1.875 1.701 1.981 1.997 1.214 -1.298 3.257 -3.176 0.199 0.814 4.123 -4.852 -3.095 2.742 -2.927 -0.034 -2.829 1.874 2.970 34.654  Prob. 0.0000 0.9701 0.0448 0.1956 0.0380 0.0226 0.0007 0.1379 0.0000 0.0000 0.0000 0.0000 0.0608 0.0890 0.0475 0.0458 0.2248 0.1944 0.0011 0.0015 0.8424 0.4158 0.0000 0.0000 0.0020 0.0061 0.0034 0.9730 0.0047 0.0609 0.0030 0.0000  Var. Mean  Var. st. dev.  0.0633 0.2671 0.2119 0.1562 0.1948 0.2429 35.5010 4.5561 21.0856 4.3567 105.3350 8.1259 71.8521 20.1507 -3.7796 22.9461 2.0170 1.0089 1.0362 0.0286 0.0505 0.0395 1.1882 8.6394 2.3558 -0.8329 0.5009 0.2612 0.2516 1.9112  0.2436 0.4426 0.4088 0.3631 0.3961 0.4289 7.3800 0.5725 5.4770 9.2949 363.3731 2.4168 37.6225 14.9848 3.8661 16.2950 0.2650 0.4176 0.5416 0.1666 0.2190 0.1949 2.1125 15.7671 4.3273 2.2418 0.8934 0.4978 0.4917 0.4661  Economic Valuation Of Critical Habitat Closures, Berman et al.  99  Rockfish, standard CPUE (cont.) ML Estimates of Selection Model Maximum Likelihood Estimates Log-Likelihood.............. -5380.0 LHS is CENSORED. Tobit Model fit by MLE. FIRST 30 estimates are probit equation. N(0,1) used for significance levels. Variable Coefficient Std. Error Constant -56.474 4.850 JUN 0.22905 0.4983 JUL 1.2618 0.6869 AUG 1.2949 1.002 SEP 2.4800 1.277 OCT 3.3829 1.587 GOA 5.2336 2.035 TIMELDEP -0.10408 0.6720E-01 LDEPTH 21.260 1.869 L2DEP -1.8490 0.1647 SLOPE 0.82802E-01 0.1053E-01 SLOPE2 -0.96891E-03 0.2243E-03 SST -0.22583 0.9279E-01 SST2 0.14440E-01 0.6475E-02 SSTSLOPE 0.35699E-02 0.2713E-02 SSH 0.30588E-01 0.1629E-01 SSHSLOPE 0.26823E-02 0.2115E-02 MWIND -0.33975 0.4520 MCHLA 0.30897 0.1708 MCHLA1 -0.33693 0.1063 DWIND 0.17681 0.1896 DCHLA 0.24109 0.1746 DCHLA1 0.64309 0.1702 GLDEPTH -0.96905 0.2408 GTIMELDE -0.49021E-01 0.2095E-01 GSST 0.19512 0.6870E-01 GSSH -0.65231E-01 0.2516E-01 GMWIND 0.53487E-01 0.6346 GMCHLA -0.44256 0.2267 GMCHLA1 0.30180 0.2029 SIGMA(1) 0.99435 0.2637E-01 RHO(1,2) -0.36229E-01 0.1442  t-ratio -11.644 0.460 1.837 1.292 1.942 2.132 2.572 -1.549 11.377 -11.226 7.863 -4.319 -2.434 2.230 1.316 1.877 1.268 -0.752 1.809 -3.171 0.932 1.381 3.778 -4.025 -2.340 2.840 -2.593 0.084 -1.952 1.487 37.712 -0.251  Prob. 0.0000 0.6458 0.0662 0.1963 0.0521 0.0330 0.0101 0.1215 0.0000 0.0000 0.0000 0.0000 0.0149 0.0257 0.1882 0.0605 0.2047 0.4523 0.0705 0.0015 0.3512 0.1674 0.0002 0.0001 0.0193 0.0045 0.0095 0.9328 0.0509 0.1369 0.0000 0.8016  Var. Mean  Var. st. dev.  100  Economic Valuation Of Critical Habitat Closures, Berman et al.  Flatfish, standard CPUE Sample Selection Model Two stage least squares regression. Dep. Variable LOGFLATS Observations 2084 Weights ONE Mean of LHS 0.1443222E+01 Std.Dev of LHS 0.7989369E+00 StdDev of resid. 0.6750596E+00 Sum of squares 0.9355633E+03 R-squared 0.2857214E+00 Adj. R-squared 0.2752838E+00 F[ 30, 2053] 0.2737428E+02 Prob value 0.3217295E-13 Log-likelihood -0.2122535E+04 Restr.(b=0) Log-l -0.2488766E+04 Amemiya Pr. Criter. 0.4624842E+00 Akaike Info.Crit. 0.2066732E+01 Standard error corrected for selection..... 0.67523 Correlation of disturbance in regression and Selection Criterion (Rho).............. -0.26454E-01 N(0,1) used for significance levels. Variable Coefficient Std. Error t-ratio Prob. Var. Mean Var. st. dev. Constant 8.5247 0.8862 9.619 0.0000 JUN -0.28334 0.1729 -1.639 0.1012 0.0614 0.2402 JUL -0.42274 0.2512 -1.683 0.0924 0.2692 0.4436 AUG -0.52272 0.3524 -1.483 0.1380 0.2116 0.4085 SEP -0.68092 0.4506 -1.511 0.1308 0.1569 0.3638 OCT -0.85019 0.5471 -1.554 0.1202 0.1934 0.3950 GOA -0.88905 0.8652 -1.028 0.3042 0.2404 0.4274 TIMELDEP 0.66403E-01 0.2414E-01 2.751 0.0059 35.5014 7.3817 LDEPTH -3.3393 0.3685 -9.061 0.0000 4.5578 0.5740 L2DEP 0.32054 0.3777E-01 8.486 0.0000 21.1029 5.4919 SLOPE -0.76981E-01 0.6136E-02 -12.546 0.0000 4.3812 9.3259 SLOPE2 0.10598E-02 0.1371E-03 7.731 0.0000 106.1247 364.6537 SST 0.10518E-01 0.4612E-01 0.228 0.8196 8.1178 2.4189 SST2 -0.17401E-02 0.3424E-02 -0.508 0.6113 71.7304 37.6320 SSTSLOPE 0.19921E-02 0.1076E-02 1.851 0.0641 20.1822 15.0117 SSH 0.30263E-01 0.7218E-02 4.193 0.0000 -3.7809 3.8750 SSHSLOPE 0.46397E-03 0.1020E-02 0.455 0.6492 22.9680 16.3217 MWIND -0.12633 0.1653 -0.764 0.4448 2.0168 0.2648 MCHLA 0.72674E-01 0.4883E-01 1.488 0.1367 1.0076 0.4184 MCHLA1 0.32083E-01 0.3426E-01 0.936 0.3491 1.0345 0.5422 DWIND -0.43080E-01 0.1009 -0.427 0.6695 0.0288 0.1673 DCHLA 0.19638 0.7851E-01 2.501 0.0124 0.0509 0.2198 DCHLA1 -0.24133 0.8384E-01 -2.878 0.0040 0.0398 0.1956 GLDEPTH 0.43883 0.9601E-01 4.571 0.0000 1.1783 2.1088 GTIMELDE -0.56250E-01 0.1005E-01 -5.598 0.0000 8.5557 15.7173 GSST -0.38102E-01 0.3501E-01 -1.088 0.2764 2.3308 4.3117 GSSH 0.29712E-01 0.1263E-01 2.353 0.0186 -0.8258 2.2383 GMWIND 0.84513 0.2942 2.872 0.0041 0.4961 0.8907 GMCHLA -0.21575 0.1022 -2.111 0.0348 0.2574 0.4944 GMCHLA1 -0.18307 0.9431E-01 -1.941 0.0523 0.2479 0.4876 IMR7 -0.17862E-01 0.5775E-01 -0.309 0.7571 1.1541 0.4255  Economic Valuation Of Critical Habitat Closures, Berman et al.  101  Flatfish, standard CPUE (cont.) Limited Dependent Variable Model - CENSORED Maximum Likelihood Estimates Log-Likelihood.............. -2189.2 Threshold values for the model: Lower N(0,1) used for significance levels. Variable Coefficient Std. Error Constant 10.023 0.9941 JUN -0.17792 0.1766 JUL -0.34136 0.2578 AUG -0.37872 0.3620 SEP -0.49852 0.4633 OCT -0.64007 0.5624 GOA -0.89924 0.8867 TIMELDEP 0.53463E-01 0.2489E-01 LDEPTH -3.8565 0.3947 L2DEP 0.38824 0.4089E-01 SLOPE -0.83725E-01 0.6378E-02 SLOPE2 0.11046E-02 0.1426E-03 SST -0.61101E-02 0.4745E-01 SST2 -0.10966E-02 0.3518E-02 SSTSLOPE 0.16883E-02 0.1102E-02 SSH 0.30473E-01 0.7427E-02 SSHSLOPE 0.43550E-03 0.1040E-02 MWIND -0.45419E-01 0.1675 MCHLA 0.36737E-01 0.4934E-01 MCHLA1 0.19910E-01 0.3539E-01 DWIND 0.44944E-01 0.1032 DCHLA 0.23501 0.8068E-01 DCHLA1 -0.26924 0.8763E-01 GLDEPTH 0.43692 0.9853E-01 GTIMELDE -0.51209E-01 0.1033E-01 GSST -0.40244E-01 0.3596E-01 GSSH 0.31127E-01 0.1292E-01 GMWIND 0.79485 0.3017 GMCHLA -0.18950 0.1047 GMCHLA1 -0.20333 0.9659E-01 IMR7 -0.14461 0.5784E-01 Sigma 0.69039 0.1092E-01  regression 0.0000  Upper  **********  t-ratio 10.083 -1.007 -1.324 -1.046 -1.076 -1.138 -1.014 2.148 -9.771 9.495 -13.128 7.748 -0.129 -0.312 1.532 4.103 0.419 -0.271 0.745 0.563 0.435 2.913 -3.072 4.434 -4.959 -1.119 2.410 2.635 -1.810 -2.105 -2.500 63.215  Prob. 0.0000 0.3137 0.1854 0.2954 0.2819 0.2550 0.3105 0.0317 0.0000 0.0000 0.0000 0.0000 0.8975 0.7553 0.1254 0.0000 0.6755 0.7862 0.4565 0.5737 0.6633 0.0036 0.0021 0.0000 0.0000 0.2631 0.0160 0.0084 0.0703 0.0353 0.0124 0.0000  Var. Mean  Var. st. dev.  0.0614 0.2695 0.2115 0.1568 0.1933 0.2408 35.5004 4.5579 21.1036 4.3791 106.0738 8.1203 71.7806 20.1982 -3.7812 22.9797 2.0167 1.0076 1.0345 0.0288 0.0508 0.0398 1.1800 8.5676 2.3361 -0.8275 0.4967 0.2578 0.2482 1.9115  0.2401 0.4438 0.4085 0.3637 0.3950 0.4277 7.3800 0.5739 5.4907 9.3241 364.5736 2.4210 37.6925 15.0259 3.8741 16.3266 0.2648 0.4183 0.5421 0.1672 0.2197 0.1956 2.1098 15.7228 4.3173 2.2392 0.8909 0.4945 0.4878 0.4670  Economic Valuation Of Critical Habitat Closures, Berman et al.  102  Flatfish, standard CPUE (cont.) ML Estimates of Selection Model Maximum Likelihood Estimates Log-Likelihood.............. -6509.3 LHS is CENSORED. Tobit Model fit by MLE. FIRST 30 estimates are probit equation. N(0,1) used for significance levels. Variable Coefficient Std. Error Constant 8.8227 1.188 JUN -0.23979 0.1790 JUL -0.35902 0.2652 AUG -0.41255 0.3767 SEP -0.54322 0.4816 OCT -0.67442 0.5914 GOA -0.83326 0.8942 TIMELDEP 0.58686E-01 0.2602E-01 LDEPTH -3.4442 0.5165 L2DEP 0.33991 0.4794E-01 SLOPE -0.81097E-01 0.6405E-02 SLOPE2 0.10962E-02 0.1424E-03 SST 0.14665E-02 0.6364E-01 SST2 -0.12020E-02 0.4484E-02 SSTSLOPE 0.20152E-02 0.1043E-02 SSH 0.30362E-01 0.6908E-02 SSHSLOPE 0.43279E-03 0.9993E-03 MWIND -0.12628 0.1637 MCHLA 0.71385E-01 0.5448E-01 MCHLA1 0.35210E-01 0.3798E-01 DWIND -0.18057E-01 0.9320E-01 DCHLA 0.20191 0.8879E-01 DCHLA1 -0.28890 0.9005E-01 GLDEPTH 0.43885 0.9877E-01 GTIMELDE -0.54051E-01 0.9655E-02 GSST -0.43807E-01 0.3895E-01 GSSH 0.30922E-01 0.1132E-01 GMWIND 0.81385 0.3022 GMCHLA -0.20505 0.9217E-01 GMCHLA1 -0.20576 0.7758E-01 SIGMA(1) 0.69119 0.1175E-01 RHO(1,2) -0.44590E-01 0.9834E-01  t-ratio 7.425 -1.339 -1.354 -1.095 -1.128 -1.140 -0.932 2.256 -6.668 7.090 -12.661 7.696 0.023 -0.268 1.932 4.395 0.433 -0.771 1.310 0.927 -0.194 2.274 -3.208 4.443 -5.598 -1.125 2.733 2.693 -2.225 -2.652 58.815 -0.453  Prob. 0.0000 0.1805 0.1758 0.2735 0.2593 0.2542 0.3514 0.0241 0.0000 0.0000 0.0000 0.0000 0.9816 0.7887 0.0533 0.0000 0.6650 0.4405 0.1901 0.3539 0.8464 0.0230 0.0013 0.0000 0.0000 0.2607 0.0063 0.0071 0.0261 0.0080 0.0000 0.6502  Var. Mean  Var. st. dev.  

Cite

Citation Scheme:

    

Usage Statistics

Country Views Downloads
China 590 14
United States 15 0
India 10 0
United Kingdom 4 0
Brazil 2 0
Russia 2 0
Vietnam 2 0
Hong Kong 2 0
Taiwan 1 0
Indonesia 1 0
Hashemite Kingdom of Jordan 1 0
Germany 1 0
Honduras 1 0
City Views Downloads
Hangzhou 318 0
Beijing 59 0
Unknown 39 1
Shanghai 28 0
Ürümqi 23 0
Guangzhou 18 0
Kunming 18 0
Jinan 15 0
Changsha 14 0
Nanjing 11 0
Harbin 11 0
Guiyang 8 0
Changchun 8 0

{[{ mDataHeader[type] }]} {[{ month[type] }]} {[{ tData[type] }]}
Download Stats

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.37052.1-0074746/manifest

Comment

Related Items