Analysis Enhancing the reliability of benefit transfer over heterogeneous sites: A meta-analysis of international coral reef values Luz M. Londoño a, 1 , Robert J. Johnston b, c, ⁎ a Research Line on Economic Valuation, Instituto de Investigaciones Marinas y Costeras — INVEMAR, Cerro Punta Betín, Santa Marta, Colombia b George Perkins Marsh Institute, Clark University, 950 Main St., Worcester, MA 01610, United States c Department of Economics, Clark University, 950 Main St., Worcester, MA 01610, United States abstract article info Article history: Received 3 August 2011 Received in revised form 23 December 2011 Accepted 27 March 2012 Available online 24 April 2012 Keywords: Coral reefs Meta-analysis Multilevel models Benefit transfer Meta-regression Valuation Willingness to pay We estimate a meta-analysis of willingness to pay for tropical coral reef recreation and evaluate its potential for international benefit transfer. The goal is improved value surface estimation and benefit transfer reliabil- ity. We compare model results to those of Brander, L., P. van Beukering and H. Cesar. 2007. The recreational value of coral reefs: A meta-analysis. Ecological Economics, 63(1), to our knowledge the only prior published meta-analysis of coral reef values. We seek to improve upon this prior model through (1) stricter attendance to methodological guidance in the meta-analysis literature, (2) greater attention to metadata uniformity, and (3) supplementation of primary study metadata with additional information obtained through secondary sources, such as information on reef characteristics from international coral reef databases. The estimated models provide value surface insights unavailable elsewhere and improve benefit transfer reliability. Results also highlight challenges in benefit transfer across heterogeneous sites and provide insight into the relevance of welfare consistency for meta-analysis. While the analysis suggests that substantial improvements in trans- fer reliability may be achieved through closer adherence to guidance from the meta-analysis literature, resulting transfers may still be subject to considerable errors. © 2012 Elsevier B.V. All rights reserved. 1. Introduction Tropical coral reefs are among the most biologically productive and diverse ecosystems on earth. These systems provide a wide range of human benefits, including those derived through fisheries, recreation, carbon storage, coastal protection and other ecosystem services (Brander et al., 2007; Cesar, 2000). Yet their rapid decline is being reported worldwide, with differentiated effects across regions (Salvat, 1992; Wilkinson, 2004, 2008). Anthropogenic threats to coral reefs include unsustainable fishing practices, coastal develop- ment and pollution. During the last decade climate change and bleaching have also emerged as threats, increasing worldwide con- cern for the sustainability of these valued systems (Wilkinson, 2008). This concern has lead to growing interest in quantifying economic benefits provided by reef ecosystems. Welfare estimates are often sought for advocacy and policymaking, for example by those seeking to quantify the benefits of conservation programs and assess tradeoffs associated with alternative reef uses (Brander et al., 2007; Cesar, 2000). Despite recent emergence of a coral reef valuation literature, however, site specific estimates are often unavailable. Reasons in- clude time and budget constraints as well as technical challenges of conducting research in areas where coral reefs are located (Brander et al., 2007). As a result, government agencies, nonprofit organiza- tions and others often seek to gain insight into the benefits of coral reefs using benefit transfer, defined as the use of results from extant primary research to predict welfare estimates for policy sites at which primary valuation estimates are unavailable (Johnston and Rosenberger, 2010). Within this context, the increasing need for reliable transfer ap- proaches is noted frequently (Brander et al., 2007; Spurgeon, 2001). Transfer reliability is the empirical accuracy of a benefit transfer, characterized by the magnitude of generalization or transfer error. More reliable benefit transfers have smaller transfer errors. In most policy applications of benefit transfer, primary study estimates are unavailable, hence reliability is unknown (i.e., benefit transfer is gen- erally required only when high quality, site specific primary studies are unavailable). However, evaluations of transfer reliability in test cases where primary study estimates are available can provide insight into the types of errors that might be expected in actual transfers (Johnston and Rosenberger, 2010; Rosenberger and Stanley, 2006). In such cases, reliability is quantified using convergent validity tests that compare a transferred welfare estimate to an available primary study estimate for a particular site. Ecological Economics 78 (2012) 80–89 ⁎ Corresponding author at: George Perkins Marsh Institute, Clark University, 950 Main St., Worcester, MA 01610, United States. Tel.: +1 508 751 4619; fax: +1 508 751 4600. E-mail addresses: lmlondono@invemar.org.co (L.M. Londoño), rjohnston@clarku.edu (R.J. Johnston). 1 Tel.: +57 5 432 8600; fax: +57 5 432 8682. 0921-8009/$ – see front matter © 2012 Elsevier B.V. All rights reserved. doi:10.1016/j.ecolecon.2012.03.016 Contents lists available at SciVerse ScienceDirect Ecological Economics journal homepage: www.elsevier.com/locate/ecolecon