EVALUATING SUPPLY-SIDE AND DEMAND-SIDE SHOCKS FOR FISHERIES: A COMPUTABLE GENERAL EQUILIBRIUM (CGE) MODEL FOR ALASKA

Abstract

This study used computable general equilibrium (CGE) models to investigate the economic effects of three exogenous shocks to Alaska fisheries: (1) reduction in pollock allowable catch (TAC); (2) increase in fuel price; and (3) reduction in demand for seafood. Two different model versions, ‘Keynesian’ and ‘neoclassical’, were used to estimate impacts on endogenous output, employment, value added, and household income. By using a CGE model, this study overcomes the limitations of fixed-price models (such as input–output models) including (1) inability to calculate welfare effects due to fixed prices; and (2) difficulty of addressing supply-side shocks. There are currently few examples of CGE studies addressing fisheries issues appearing in the literature. Among those, this study is unique in that it uses a relatively disaggregated sector scheme and examines both supply-side and demand-side shocks.

Keywords:

• Computable general equilibrium
• Alaska
• Fisheries impacts

Acknowledgments

The authors would like to acknowledge funding and support from the National Marine Fisheries Service Alaska Fisheries Science Center.

Notes

¹ We use an industry-by-commodity ‘make’ matrix as well as a traditional commodity-by-industry ‘use’ matrix in our CGE model. This means that each industry can produce multiple commodities, and vice versa. There are two seafood processing industries, ‘pollock-processing’ and ‘other species-processing’. Both produce the same commodity called ‘seafood’. The number of commodities is therefore one less than the number of industries, and output and price for ‘seafood’ is determined by the interaction between unitary demand for the commodity and supply produced by the two processing sectors.

² Houston et al. (1997) constructed a CGE model of the Oregon coastal region, which is an aggregation of all coastal counties in the state of Oregon, USA.

³ A discussion of this type of trade model specification, known as the ‘Armington assumption’ and its use in CGE models is found in Shoven and Whalley (1984). Most importantly for our purposes, it facilitates model tractability when the same type of good or service is simultaneously produced and used locally (or exported) as well as imported.

⁴ The Alaska CGE model used in this study is a static model, meaning that pre- and post-shock general equilibrium scenarios are estimated and compared directly; there is no estimation or examination of intermediate steps along an optimal time path between the two equilibria. Two model versions were used in this paper, one in which there is no interregional movement of factors of production (Neoclassical version), and one in which labor is allowed to move between regions (Keynesian version). Although there is no clear evidence on how long it takes for an economy to reach a new equilibrium following a shock, the results from the two model variants provide short-run (1–2 years, Neoclassical model) to intermediate-run (2–5 years, Keynesian model) perspectives. Generally, these versions of regional CGE models are appropriate for analysis of large regions. In contrast, fixed price models are generally more useful for estimating relatively long-run effects in small regions or other regions where interegional factor mobility is not constrained.

⁵ Another possible specification of interregional labor mobility is to use a labor migration elasticity. Regional CGE modelers often use a labor migration function of the following form: where LMIG is net labor in-migration; LSB is labor supply in base year; λ is labor migration elasticity; W is average wage rate in the region of interest; and W _ROW is the average wage rate in the rest of the world. Plaut (1981) estimated a labor migration elasticity of 0.92. If this specification were used, the impacts on wage rate and labor migration would be somewhere between the levels calculated by the neoclassical model (in which the migration elasticity is zero) and the levels calculated by the Keynesian model (in which the elasticity is infinity).

⁶ The Alaska Fisheries Economic Assessment Model (FEAM) was constructed in the 1990s for estimating impacts of changes in fish landings. FEAM uses IMPLAN response coefficients and survey-based estimates of harvester and processor input purchasing patterns to estimate impacts on coastal communities of changes to status quo fisheries.

⁷ Interviews were conducted as part of an informal data collection process by the authors and their contractors. For details see The Research Group (2007).

⁸ Although it is acknowledged that more advanced techniques such as RAS and cross-entropy techniques are available, wherever possible manual adjustments were made to those accounts that are used by IMPLAN as residual balancing items. As a result, distortions were likely minimized.

⁹ The fuel price in this study is the price of refined petroleum, and is exogenous across the economy.

¹⁰ Although, in the Keynesian CGE, total output in the combined non-seafood sectors decreases by only 0.3% as a result of higher fuel price, output of some individual non-seafood industries decreases by a much larger percentage. For example, outputs in the ‘Other Manufacturing’ and ‘Transportation’ sectors, which depend heavily on fuel, decrease by 6.5% and 2.8%, respectively.

¹¹ We developed a fixed-price SAM model from the CGE dataset, and used it to simulate the effects of a reduction in seafood exports by the same amount implied in experiment 3 in the Keynesian version of the CGE model ($242.4 million). Total seafood sector output decreases by 20.2% more with the SAM model than with the CGE model (–$335.4 million versus –$279.1 million), while the total output in the Alaska economy decreases by 61.2% more with the SAM model than with the CGE model (–$557.54 million versus –$345.9 million).

¹² Since the parameter being examined is the elasticity of substitution between labor and capital, relatively small differences in output can mask much greater swings in underlying factor utilization levels.