. G6 (£009: Dec. ) 5 Library ( t IANNINI • FOUNDATIO N OF AGRICULTURAL ECONOMICS UNIVERSITY OF CALIFORNIA An Economic Evaluation of the Hass Avocado Promotion Order's First Five Years Hoy F. Carman, Lan Li, and Richard J. Sexton (^iannini Foundation Research Report 351 December 2009 University of California Agriculture and Natural Resources THE AUTHORS Hoy F. Carman is professor emeritus and Richard J. Sexton is professor, Department of Agricultural and Resource Economics, University of California, Davis. Lan Li is research associate, National Institute for Commodity Promotion Research, Cornell University. Contact: Richard Sexton, rich@primal.ucdavis.edu, 530.752.4428. U£ ^IR This publication has been anonymously peer-reviewed PEER ' or tec hnical accuracy by University of California REVIEWED scientists and other qualified professionals. ©2009 by the Regents of the University of California Division of Natural Resources All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording, or otherwise, without the written permission of the publisher and the authors. To simplify information, trade names of products have been used. No endorsement of named or illustrated products is intended, nor is criticism implied of similar products that are not mentioned or illustrated. An Economic Evaluation of the Hass Avocado Promotion Order's First Five Years TABLE OF CONTENTS 1. Introduction 1 2. Avocado Promotion Programs 2 2.1. California Avocado Commission Programs 2 2.2. Chilean and Mexican Avocado Importer Association Programs 3 2.3. Hass Avocado Board Programs 4 3. Avocado Consumption in the United States 5 4. Modeling Annual Demand for Avocados 7 4.1. Previous Studies 7 4.2. Econometric Models of the Annual Demand for Avocados 7 4.3. Preliminary Data Analysis 9 4.4. Structural Breaks in Per Capita Consumption 11 4.5. Estimated Annual Demand Relationships 12 4.6. Diagnostic Checks of Annual Demand Models 14 4.7. Two-Stage Least Squares Estimation 14 4.8. Summary 16 5. Benefit-Cost Analysis 18 5.1. Benefit-Cost Analysis in Promotion-Evaluation Studies 18 6. Demand Analysis at the Retail Level 22 6.1. The Data 22 6.2. The Econometric Models 23 6.3. Results 25 6.4. The Effects of the California Avocado Commission's Promotions on Retail and Shipping-Point Prices 31 7. Evaluation of the Hass Avocado Board's Network Marketing Center Program 33 7.1. Variability of Prices and Quantities over Time 33 7.2. Costs of the Hass Avocado Board's Information Program 34 7.3. Estimated Benefits from the Information Program 35 8. Conclusions 37 References 39 Giannini Foundation Research Report 351 FIGURES 1. Annual U.S. Per Capita Avocado Consumption by Source, 1980-2007 6 2. Annual Per Capita Avocado Consumption, U.S. Per Capita Disposable Income, and the Percentage of the Population That Is Hispanic, 1962-2007 10 3. Annual Per Capita Avocado Consumption, FOB Price, and Promotion Expenditure, 1962-2007 10 4. Avocado Supply, Imports, and Domestic Production in the United States, 1962-2007 17 5. Avocado Promotion Simulation Model 19 ii An Economic Evaluation of the Hass Avocado Promotion Order's Eirst Five Years TABLES 1 . U.S. Avocado Promotion Expenditures in Dollars by Organization, 2003-2007 3 2. Variable Definitions and Summary Statistics 8 3. Correlation Coefficients for the Demand Model 11 4. Estimated Annual Demand Models: Ordinary Least Squares 13 5. The First-Stage Regression to Predict California FOB Price 15 6. Estimated Annual Demand Models: Two-Stage Least Squares 16 7. Simulation Model Results 21 8. Summary Statistics for the Disaggregate Model 23 9. Estimation Results for the Retail Sales Model: Within Model 26 10. The Effects of Promotion on Retail Sales from Panel I: 2003/04 28 11. The Effects of Promotion on Retail Sales from Panel II: 2007/08 29 12. The Effects of California Avocado Commission Promotions on Retail Price and Shipping-Point Price 32 13. Standard Deviation of Weekly California and Total Avocado Shipments, 2003-2007 33 14. Annual and Total Costs of the Hass Avocado Board's Information Programs by Cost Category, 2003-2007 34 15. Estimated Total Annual Changes in Gross Margins for Hass Avocados, Average Shipments, Standard Deviations of Price, and Average Prices for 2003-2007 35 SER.REC LIBRARY MAR 1 6 2010 U.C DAVIS Giannini Foundation Research Report 351 An Economic Evaluation of the Hass Avocado Promotion Order's First Five Years 1. INTRODUCTION The U.S. avocado industry has evolved from an emphasis on seasonal domestic production of a mix of avocado varieties to year-round availability of domestic and imported Hass avoca- dos. California avocado producers, who account for approximately 90% of U.S. avocado production and essentially all U.S. Hass avocado production, have funded promotional programs for avocados since 1961. With few imports of avocados prior to the early 1990s, the benefits from these demand-enhancing programs flowed directly to California producers. Imports of avocados into the United States have increased steadily since then, resulting in a free-rider problem that led ultimately to creation of the Hass Avocado Promotion, Research, and Information Act of 2000 that was signed into law by President Clin- ton on October 23, 2000. This act established the authorizing platform and timetable for creation of the Hass Avocado Promotion, Research, and Information Order (HAPRIO), which was approved in a referendum of producers and importers with 86.6% support on July 29, 2002. This study evaluates the promotion activities conducted by the Hass Avocado Board (HAB) dur- ing its first five years of operation. The evaluation analyzes the impacts of the expenditures and the overall returns accruing to Hass avocado producers from all promotion programs. For some of the statisti- cal methods employed in this evaluation, a five-year period provides insufficient data. In these situations, we evaluate the entire history of avocado promotion from the beginning of organized efforts in California in 1961 to the present. Aside from providing new information on the effectiveness of promotion for an important California specialty crop, several features of the study distinguish it from predecessor works. 1 First, HAB is unique in that it involves two international importer associa- tions, along with a domestic producer board, making evaluation of the effectiveness of this innovative alli- ance a unique undertaking. Second, the study involves analysis of both aggre- gate annual time-series data, as has been common for perennial crops (Kaiser et al. 2005), and disaggregate scanner data collected at the retail level. These data enable the study to investigate the question of whether retailers, through their pricing strategies, capture a portion of the benefits of promotion through higher prices. Finally, the study develops a benchmark method- ology to evaluate the innovative information-sharing program implemented by HAB. By widely sharing information among market participants on harvests and shipments, HAB hopes to smooth flows of prod- uct to markets, prevent occurrences of shortages and surpluses, and stabilize prices at both free-on- board (FOB) and retail levels. We do find evidence of increased price stability in the presence of the program and translate that impact into reductions in the mar- keting margin based on the results for avocado price transmission provided in Li (2007). Section 2 of the report discusses the major mar- keting programs conducted under the auspices of HAB. Section 3 provides an overview of trends in U.S. consumption of avocados, while section 4 contains a detailed analysis of annual demand for avocados in the United States with the goal of determining the impact that promotions have had on avocado demand. Section 5 introduces and implements a simulation framework to estimate the impact of avo- cado promotion on grower prices and incomes and on consumption of avocados based on the results of the demand analysis. Section 6 provides an analysis of avocado demand and the impact of promotion based on retail scanner data. Section 7 presents the analysis of the impacts of HAB's information-sharing and dis- semination program. Finally, section 8 presents brief concluding remarks. Commodity promotion evaluation studies have been an important applied research topic in agricultural economics. The recent book by Kaiser et al. (2005) summarizes much of this literature. I Giannini Foundation Research Report 351 2. AVOCADO PROMOTION PROGRAMS The HAPRIO took effect on September 9, 2002, with program assessments beginning onjanu- ary 2, 2003. The twelve-member board that administers the program under U.S. Department of Agriculture (USDA) supervision consists of seven domestic producers and five importers. Appointment of the first HAB members on February 12, 2003, initiated activities under the HAPRIO. The manda- tory assessment rate is 2.5$ per pound for all Hass avocados sold in the United States and the maximum permitted assessment is 5.0 Number of retail accounts 90 78 Number of clusters (account size) 147 142 Number of observations 13,886 14,166 Maximum number of observations per cluster 118 104 Average number of observations per cluster 94 53 23 Giannini Foundation Research Report 351 but include only expenditures on radio promotions during 2007/08. Alternatively, expenditures on radio and outdoor promotions can be introduced as separate explanatory variables in Panel I to estimate differential promotion effects for different media types on avocado demand. Advertising may have a dynamic effect on demand, increasing sales in the current period and into the future. Such effects may be captured by specifying a distributed lag structure for advertising expenditures (Erdem and Keane 1996; Ackerberg 2001, 2003). How- ever, the nature of CACs promotions made it difficult to apply such models because the weekly expenditure in each treatment market during the promotion peri- ods was essentially constant. As a result, the data do not provide sufficient variations between periods to estimate a distributed lag structure. We first present a framework for program evalua- tion based on a binary treatment variable (the presence or absence of a promotion campaign) and then extend this method to the amount of promotion expenditure, which is a continuous treatment variable. When the promotion variable is binary (i.e., advertising is pres- ent in the market (D = 1) or is absent (D = 0)), the approach of Difference in Difference (DID) is used. The fact that CAC selected a subset of markets for its promotion programs from among the total group on which scanner data were available enables us to construct both treatment and control groups for the program evaluation. The DID approach estimates counterfactual outcomes for the retail accounts in the selected markets that were exposed to CAC's promo- tion programs. The DID framework for identifying the "treatment effects" of CAC's promotions on retail sales can be presented by the following linear model: q(a,t) = 8(0 + 11(a) + yD(a,t) + u(a,t) where q(a,t) denotes retail sales of avocados at retail account a at time t. Let the pretreatment period, ( = 0, be the period when there was no promotion and let the treatment period, t = 1, be the period when CAC conducted its promotions. D(a,t) denotes whether or not a retail account was exposed to CAC's promo- tions at time t. Suppose that only q(a,t) and D(a,t) are observed. We refer to retail accounts that were exposed to CAC's promotion programs (when D(ci,l) = 1) as "treated" and those that were not exposed to the promotions (when D(a,l) = 0) as "controls." The term D(a,0) equals zero for both the treated and the control group because there was no promotion at t = 0. The coefficient \\i represents the treatment effect of CAC's promotion programs, 5(t) denotes the time-specific component, n(a) represents the account-specific effects, and u(a,0 is the individual transitory error term with a zero mean at both t = 0 and t = 1. The advantage of the panel data set is that it enables us to control idiosyncratic characteristics of individual retailers or markets by use of fixed effects. Under the assumption that CAC's selection for treatment is not correlated with the error term, 22 we can obtain the difference in expected retail sales with and without CAC's promotions for the retail accounts in the treated and control markets as: E[q(a,l)|D(a,l) = 1] -E[q(a,0) | D(a,l) = 1] = E[q(a,l)-q(a,Q)\D(a,l)= 1] = [5(1) -6(0)] + h(a) -n(«)] + v[D(a,l)-D(a,0)] = 5(1) -5(0) + y E[i,uuuj U.UJ.T -0. 139 (0.080) Christmas /New Year 7.06*** (1.82) 6. 17*** (1.30) Super Bowl 17 59*** (3.86) 15 94*** (4.52) Valentine's/Presidents' Day 0.73 (1.21) -0.43 (2.10) Oscar Awards 1.29 (1.25) -4.53* (2.33) Cinco de Mayo 4.52** (2.11) 11.12*** (4.03) Easter 1.52 (1.40) -0.74 (1.22) Mothers' Day -1.20 (1.50) -0.44 (1.87) ivic 1 1 IU1 1 cx I L^tiy 1 70* (i.-tj ; 1. It (2.U3) Fathers' Day -3.52 T3 241 n (Si 1 1 87^ (,1.0/ ) Independence Day (1.35) 5 12*** (1.36) Labor Day 5 ^q*** (1.70) (0.88) Thanksgiving -1.51 (2.93) -7.42*** (1.88) 2002 (base) 0.0 0.0 2003 0.97 (1 41) 2004 4.78** (2 25) 0.0 0.0 ?oos 2.88 (2.15) January -1.53 (1.18) -0.59 (2.01) February -1.60 (1.34) 3.30 (2.67) March -0.64 (1.44) 3.69* (1.95) April -0.29 (1.50) 1.71 (1.82) \ 1 1 \ ' -> (SO*** (1.39) 8.62 (2.73) June 7.74** (3 16) V.-'. iw / 5.26** O 301 July 0.00 d 411 VJ ._)T August 1.17 d 501 ^ 1 40 "1 (.i.Ty; September 4.34*** (1.63) 0.06 (1.12) November 1.57 (1.75) -0.45 (1.68) December -1.21 (1-75) -6.10*** (1.50) Constant -4.11*** (1.35) 90.06 (9.78) R 2 0.108 0.062 Number of observations 15,360 13,838 Maximum no. of observations per cluster 118 104 Number of clusters 147 142 Note: Asterisks denote statistical significance levels with one, two, and three stars denoting statistical significance at 90%, 95%, and 99%. respectively. 26 An Economic Evaluation of the Hass Avocado Promotion Order's First Five Years An alternative model framework included lags of retail sales of a variety of lengths but the lagged sales had no important effects either in terms of magnitude or statistical significance. These results indicate the absence of state dependence in weekly retail sales of avocados, or the absence of a consumption habit effect for avocados, as measured on a weekly basis. 26 Turning now to promotion, the measured impact of CACs promotion program on retail sales for avocados is comprised of two effects. First, CACs promotion may contribute to higher average retail sales in pro- motion markets relative to nonpromotion markets. However, the estimated coefficient of the promotion variable from the between-effects model consists ol both the effect of promotion and the effect of other unobserved factors on average retail sales. These unobserved factors, which may contribute to differ- ences in average retail sales between promotion and nonpromotion markets, need to be controlled to attain clean identification of the effect of promotion. Second, the effect of the promotion program on retail sales is measured by how much retail sales deviate from mean levels as the promotion expen- diture increases or as the promotion program is conducted in a week. This effect is estimated by the within model by controlling the difference in average retail sales between promotion and nonpro- motion markets. Clearly, the overall promotion effect includes both the effect on average retail sales and the effect on deviations of retail sales from the aver- age in promotion markets. Both the pooled and the random-effects models utilize variations in promotion both in cross-section and over time to estimate the overall promotion effect. The random-effects model is preferred to the pooled model in the presence of unobserved heterogeneity. Nevertheless, both provide valuable information. Tables 10 and 11 present the estimated effects of promotion on retail sales by the pooled, between- effects, within, and random-effects models for the data in Panel 1 and Panel II, respectively. Estimation I, which is reported in the top portion of each table, does not control unobserved factors that may contribute to the difference in average retail sales between promotion and nonpromotion markets. As a result, the estimated promotion effect from the pooled, between-effects, and random-effects models could be biased. Estima- tion II introduces a dummy variable to control the difference in average retail sales between promotion and nonpromotion markets in the pooled, between- effects, and random-effects models. Estimation III uses dummy variables for individual promotion markets to control differences in average retail sales between each promotion and nonpromotion market. 2 ' Consider first the estimates from Panel 1, which are shown in Table 10. The estimated promotion effects from the between-effects model show that average weekly retail sales for each size of avocado at retail accounts were 1,472 units greater in promo- tion markets than in nonpromotion markets for each $1,000 of weekly promotion expenditure (estimation III). The average weekly promotion expenditure was $20,419 in a promotion market. Therefore, average weekly retail sales increased by an estimated 30,057 units for each size of avocado sold at retail accounts in promotion markets. Introducing dummy variables to control unobserved factors that may contribute to differences in average retail sales between promotion and nonpromotion markets reduces the size of the estimated promotion effect significantly (compare the promotion coefficients in estimation 1 versus esti- mation III). The introduction of additional variables decreases efficiency in the estimations and increases standard errors. The increase in average retail sales in promotion markets due to promotion is positive but is not statistically significant for estimations II and III. Second, the results from the within model indicate that sales at a retail account in a promotion market increased slightly as promotion expenditure increased during promotion periods but the effect was not statistically significant. 26 The other rationale to include lagged retail sales in the model is that serial correlation may exist in retail sales. If that is the case excluding the lagged retail sales could generate omitted variable bias. This did not appear to be an issue for the estimation. The post-estimation tests show that residuals are not serially correlated and overidentification tests show that the set of variables is exogenous. 27 In other words, estimation 111 contains a separate (0,1) indicator variable to identify market-specific effects on sales (see the bottom portion of Table 10). 27 Giannini Foundation Research Report 351 Table 1 0. The Effects of Promotion on Retail Sales from Panel I: 2003/04 Pooled Between Within Random Effects Estimate Std Error Estimate Std Error Estimate Std Error Estimate Std Error Estimation I Retail price ( -0.654*** 0.128 -0.701** * 0.180 -0.508*** 0.088 -0.509*** 0.088 t - 1 0.092* 0.052 0.155*** 0.042 0.154*** 0.042 t-2 -0.003 0.064 0.135*** 0.024 0.134*** 0.024 Promotion expenditure 0.792** 0.322 2.741** 1.107 0.014 0.060 0.017 0.060 Small size -9.959 7.786 -15.059 12.019 -1.077 10.675 R 2 0.115 0.167 0.108 0.072 Root mean square error 72.811 63.634 32.482 32.482 Intraclass correlation 0.803 A 0.000 1.000 0.954 Estimation II Retail price t -0.621*** 0.116 -0.659** ' 0.181 -0.509*** 0.088 t-1 0.102** 0.051 0.155*** 0.042 t-2 0.036 0.056 0.134*** 0.024 Promotion expenditure 0.059 0.386 0.288 1.806 0.014*** 0.060 Promotion market 36.510*** 12.019 31.060* 18.128 39.006*** 11.671 Small size -9.245 7.760 -14.782 11.940 -2.000*** 10.087 R 2 0.158 0.184 0.141 Root mean square error 71.016 63.208 32.482 Intraclass correlation 0.801 A 0.000 0.954 Estimation III Retail price t -0.591*** 0.102 -0.581*" 0.170 -0.509*** 0.088 t - 1 0.116** 0.048 0.155*** 0.042 1-2 0.065 0.045 0.134*** 0.024 Promotion expenditure 0.035 0.069 1.472 3.701 0.014 0.060 Promotion market Phoenix 16.951 13.343 5.825 26.913 21.829* 12.018 Los Angeles 69.981*** 25.560 39.899 67.035 73.630*** 25.476 San Francisco 39.372* 23.723 25.232 39.694 40.201* 23.523 Atlanta -7.070 8.219 -27.905 53.288 -2.742 6.531 Portland 2.226 8.329 -2.324 29.755 3.070 7.553 Dallas 4.710 9.940 -10.090 28.089 9.278 9.572 Houston 48.944* 28.158 35.689 28.709 54.332* 30.497 San Antonio 289.623 196.875 280.932*** 38.019 296.148 201.896 Small size -4.831 7.487 -10.423 10.421 -0.834 8.631 R 2 0.364 0.433 0.355 RMSE 61.763 54.025 32.482 Intraclass correlation 0.748 A 0.000 0.946 Note: Asterisks denote statistical significance levels with one, two, and three stars denoting statistical significance at 90% 95% and 99<3i respectively. 28 An Economic Evaluation of the Hass Avocado Promotion Order's First Five Years Table 1 1 . The Effects of Promotion on Retail Sales from Panel II: 2007/08 Pooled Between Within Random Effects Estimate Std Error Estimate Std Error Estimate Std Error Estimate Std Error Estimation I Retail price f -0.759*** 0.121 -0.272** n 1 34 \J. 1 JU n l i~k U.1ZJ W. / JU A IT) t- 1 0.245** 0.097 0.252*** 0.096 0.252*** 0.096 t-2 0.083 0.058 0.096** 0.043 0.096** 0.043 Radio expenditure 1.130*** 0.278 16.419*** 2.167 -0.139 0.080 -0.134* 0.080 Small size -30.349*** 11.080 -25.143*** 9.686 -29.703*** 10.967 R 2 Root mean square error Intraclass correlation A 0.117 69.867 0.000 0.350 53.523 0.193 31.489 1.000 0.100 31.706 0.743 0.942 Estimation II rvcLaii ui iee i A 117 U. 1 W U.Z /4 U. i JO -0.736 0.122 t-i 0.254** 0.100 0.252*** 0.096 t-2 0.099* 0.058 0.096** 0.043 Radio expenditure 0.352** 0.176 16.169*** 3.608 -0.134* 0.080 Radio promotion market 55.327*** 12.019 1.378 16.013 -0.405 13.922 Small size -29.911*** 7.760 -25.206** 9.749 -29.705 10.984 R 2 Root mean square error Intraclass correlation A 0.236 64.991 0.000 0.350 53.716 0.098 31.646 0.744 0.942 Estimation III Retail price t -0.685*** 0.102 -0.174 0.123 -0.735*** 0.088 t- 1 0.290*** 0.048 0.252*** 0.042 t-2 0.163*** 0.045 0.097** 0.024 Radio expenditure -0.140 0.069 -22.684 36.705 -0.136* 0.060 Radio market Phoenix 49.559*** 17.283 93.606 71.562 50.197*** 15.955 Los Angeles 118.715*** 30.768 298.736 291.426 115.488*** 31.392 San Francisco 191.403*** 54.709 290.859* 160.341 187.585*** 55.119 Atlanta 4.303 16.164 126.789 185.952 -6.675 14.974 Portland 15.702 13.438 40.357 50.450 20.182* 11.263 Dallas 3.562 5.363 74.942 116.306 2.980 5.025 Houston 45.722** 23.012 116.805 115.186 44.471** 30.497 Seattle 17.154 15.017 51.899 62.819 22.606 14.211 Small size -22.650*** 7.558 -21.489** 8.559 -26.449*** 8.220 R 2 RMSE Intraclass correlation A 0.454 54.987 0.000 0.532 46.773 0.448 31.646 0.687 0.933 Note: Asterisks denote statistical significance levels with one, two, and three stars denoting statistical significance at 90%, 95%, and 99%, respectively. 20 Giannini Foundation Research Report 351 Third, the estimated promotion effects from the random-effects model were the same as those from the within model after effects of individual promotion markets were controlled in estimation II and III. This suggests that promotion on the whole increased aver- age retail sales for promotion markets over those in nonpromotion markets. Promotion effects from week to week in each promotion market were small. This is likely due to the promotion effects merging over time, increasing average retail sales as a whole. Therefore, the majority of the promotion effect was identified via the between-effects model rather than by the within model. 28 This conclusion is consistent with the results of Erdem and Keane (1996), who were able to apply a distributed lag structure to discern the impact of advertising on detergent sales and concluded that advertising had weak short-run effects but a strong cumulative effect in the long run. Table 11 presents the estimation results from Panel II (2007/08). First, the estimated promotion effects (estimation II) from the between-effects model show that average weekly retail sales for each size of avocado at retail accounts in promotion markets were 16,169 units greater than those in nonpromotion markets for each $ 1,000 of average weekly promotion expenditure. The estimated impact of these radio promotions is eleven times greater than the estimated impact of CAC's promotion program (which included both radio and outdoor promotion) during the 2002/03- 2003/04 crop years. The average weekly promotion expenditure was $37,360 in a promotion market. Therefore, average weekly retail sales increased by an estimated 604,474 units for each size of avocado sold at retail accounts in promotion markets. Estimation II controls unobserved factors that may contribute to differences in average retail sales between promotion and nonpromotion markets. The results show that the difference was primarily and significantly explained by the difference in average retail sales due to promotion. These results are differ- ent from those for 2003/04 (Table 10), which show that the estimated promotion effects were reduced significantly after controlling the unobserved factors that may account for differences in average sales between promotion and nonpromotion markets. Estimation III includes dummy variables for each promotion market in the data to control differences in average retail sales between promotion markets. The estimated coefficient for the radio promotion expen- diture is negative and not statistically significant in estimation III. This is mostly due to inevitable multi- collinearity problems because (1) the radio promotion expenditure is highly collinear with the set of dummy variables for promotion markets and (2) introduction of additional variables decreases estimation efficiency. Nonetheless, the estimated coefficients for promotion markets provide valuable information by identifying the markets in which average retail sales were signifi- cantly higher. Note that the estimates may represent some effect of radio programs because the promotion variable is highly correlated with the dummy variables for promotion markets. The estimates also may rep- resent the effect of other promotion programs that are not included in the model due to lack of available data. The coefficient estimates for individual promo- tion markets are markedly higher for 2007/08 than for 2003/04. The markets associated with the highest average sales in 2007/08 were Los Angeles, Atlanta, San Francisco, Houston, and Phoenix while the high- est sales in 2003/04 were San Antonio (data were not available for San Antonio during 2007/08), Los Angeles, Houston, and San Francisco. Second, the results from the within and random- effects models indicate that radio promotions had an insignificant effect on retail sales during the promo- tion periods both in terms of the magnitude of the estimated effects and their statistical significance. This may be because the effects of individual radio promotions were consolidated over time and increased average sales in the overall market and/ or because the effects of radio programs were blended with the effects of other promotion programs that were conducted but are omitted from the estimation (due to lack of avail- able data), such as programs that were conducted in the treatment cities during 2007/08 by CAIA, MHAIA, and/or HAB. " l In results not reported here but available from the authors upon request, we investigated decomposing the impact of CAC's expenditures by media type— radio advertising versus outdoor advertising. Results suggest a greater effectiveness for radio pro- motions relative to outdoor advertising but the effects were not statistically significant. 50 An Economic Evaluation of the Hass Avocado Promotion Order's First Five Years Taken together, the results suggest that radio promotion significantly increased average retail sales in promotion markets compared with average retail sales in nonpromotion markets. The estimated promo- tion effects in 2007/08 are also markedly higher than the promotion effects during 2003/04. The various models reported in Tables 10 and 11 estimated promotion effects from different perspectives but the most relevant results are those for estimation II from the between-effects model. Those effects are comprised of two elements: a significant increase in sales during the promotion period (the wave) and a sales increase stretched during and beyond the pro- motion period (level). The results from panels I and II consistently indicate that the second effect (increase in the average level) dominates. The first effect, mea- sured by the within and random-effects models, is small because (1) promotion expenditures did not vary much from week to week and/or (2) because the effects of promotion were carried over time and, hence, the promotion effects were consolidated to increase average retail sales (the average level). A comparison of the results from estimations I and II suggests that the heterogeneity between promotion and nonpromotion markets needs to be controlled and, hence, that estimation II is preferable to estimation I. The results from estimation III must be viewed with some caution due to the problem of multicollinearity. The value of estimation III is the resulting ability to see comparative or relative promotion effects between markets and between two panel periods rather than having to observe only absolute estimates. 6.4. The Effects of the California Avocado Commission's Promotions on Retail and Shipping-Point Prices The findings for retail sales for avocados suggest, on balance, that CAC's promotion program had a posi- tive effect on retail sales for avocados. We now seek to determine the impacts, if any, of the program on retail prices and prices at the shipping point. This analysis is conducted only on Panel I during 2003/04 because expenditure data were not available for both radio and outdoor programs during 2007/08. Table 12 presents the estimated effects of CAC's promotion program on the retail and shipping-point price for avocados. Both models use weekly dummy variables as time-control variables. Panel 1 in Table 12 contains the estimated impact of incremental $1,000 expenditures by CAC in the targeted market while panel 2 estimates the cumulative impact in terms of total expenditure during each of the promotion periods. If arbitrage at the shipping level is efficient between promotion and nonpromotion markets and between promotion and nonpromotion weeks or periods, CAC's promotion program should have no significant effect on shipping-point (POB) prices to alternative destination markets. Successful promotions in tar- geted markets increase demand in those markets, which should cause shippers to expand shipments to those markets relative to nonpromotion markets. This reallocation of supply between markets should continue until the shipping-point prices to all desti- nation markets are equated if arbitrage by shippers is efficient. A similar argument applies to intertem- poral arbitrage designed to have shipments in place at destination markets to coincide with promotion periods. The results for the shipping-point price of avocados shown in Table 12 reveal a small positive estimated impact of promotion on prices to target markets. Shipping-point prices to promotion mar- kets during promotion periods were 0.116C per unit higher (panel 2) than shipping-point prices during nonpromotion periods and to nonpromotion markets. However, this estimated impact is not statistically significant for either radio or outdoor promotion by CAC, which is consistent with the efficient arbitrage hypothesis. A possible concern for industry generic promotion relates to retailer responses to promotions. Successful promotions targeted to consumers increase demand for avocados. That increase in demand can be reflected in increased retail sales, higher retail prices, or a com- bination of both. To the extent that retailers increase prices and margins to capture the higher demand generated by industry promotions, their actions vitiate the effectiveness of the programs from the industry's perspective. The increase in sales that is necessary to 31 Giannini Foundation Research Report 351 increase the price received by growers and shippers will not occur. Conversely, if retailers did not capture the benefits of increases in demand for avocados due to CAC's promotion program through higher prices, we would expect the benefit to accrue at the grower- shipper level in the form of higher FOB prices. An interesting possibility supported by some research is that retailers may reduce retail prices in response to a positive demand shock for a product, most likely as a way to entice customers to the store to then purchase additional items. Evidence of lower retail prices for avocados in response to CAC's promo- tion program would mean that retailers' actions are reinforcing (instead of offsetting) the impacts of CAC's promotion. The results in Table 12 show that retail prices during promotions were 0.363%, and 99%, respectively. ^ These effects are found by multiplying the estimated coefficient on pooled promotion contained in the top hall of the table by the mean promotional expenditure in the targeted markets. 32 All Economic Evaluation of the Hass Avocado Promotion Order's First Five Years 7. EVALUATION OF THE HASS AVOCADO BOARD'S NETWORK MARKETING CENTER PROGRAM HAB conducts an active internet information program through its Network Marketing Center to share information to promote orderly marketing. As stated in its hrst annual report (2003): The primary goal behind the INFOTECH plank of HAB's Strategy is to develop "Stra- tegic Intelligence" that will enable avocado marketers to share information essential to orderly marketing throughout the full 12-month season and ameliorate seasonal transition points and concomitant market instability between sources. This initiative is designed to help ALL sellers in the U.S. market develop a much-needed framework to ensure orderly flow of fruit and market stability. The benefits from such an end state would inure to consumers, supermarket retailers and those suppliers selling Hass avocados in the U.S. (p. 11-12) All participants in the Hass avocado marketing chain have access to the HAB website (AvoHQ.com) where they are able to share harvest and shipment planning information. The 2006/07 annual report (page 3) indi- cates that HAB's technology infrastructure supported more than 2,500 users and use has continued to grow since then. This ongoing information exchange is intended to smooth shipments to major U.S. markets, prevent surplus and shortage situations, and promote stable FOB and retail pricing. first examined the standard deviation of California FOB avocado prices for the most recent ten-year period (1998-2007). While there was not an evident trend over time, the weekly standard deviation of the FOB price for the most recent five years averaged 0.2045 while the same average for the first five years was 0.2843. At the same time, the weekly standard deviation of California shipments increased from an annual average of 2,293,841 pounds for the first five years (1998-2002) to 4,303,944 pounds for the most recent five years (2003-2007). This indicates that, while the size of California shipments has become more variable, efforts to coordinate imports with Cali- fornia shipments have smoothed total weekly avocado shipments and prices during the marketing year. The most recent five-year comparison of California ship- ments with total weekly shipments (California plus imports) is shown in Table 13. While rising imports have the potential to introduce additional quantities and price variability into the U.S. market, the opposite has occurred. Imports have been timed to maintain a rather steady flow of avocados to retail markets, which tends to stabilize prices at both the FOB and the retail level. A portion of the smoothing of quantities and prices despite significantly increased imports can and should be attributed to the active HAB informa- tion program. Previous research on specialty agricultural com- modities has demonstrated that decreased price variability can benefit both producers and consumers. 7.1 . Variability of Prices and Quantities over Time Empirical evaluation of the benefits of an information program is difficult and the activities of HAB are no exception. We can, however, examine some industry statistics related to HAB's goal of an orderly flow of fruit and market stability that provide an indication of progress toward meeting program goals. We mea- sure the variability of prices and quantities over time using the standard deviation of weekly prices and quantities for California and imported avocados. We Table 1 3. Standard Deviation of Weekly California and Total Avocado Shipments, 2003-2007 Year California Total: California Plus Imports 2003 3,359,560 1,479,939 2004 5,020,240 2,693,992 2005 4,593,614 2,052,438 2006 6,399,061 3,330,162 2007 3,483,128 1,990,026 Five-year average 4,303,944 2,309,312 3 3 Glannini Foundation Research Report 351 Market conditions present in the U.S. avocado indus- try that can lead to this result are: (1) food retailers that have market power in setting their retail prices, (2) a perishable product, and (3) retail chains that purchase the product directly from grower-shippers with operations that are small relative to the chain buyers. Under these conditions, buyers can use large or temporarily large supplies to bid down shipping- point prices and increase their margins (Sexton and Zhang f 996). These same conditions can also lead to asymmetric price transmission from the producer to the retail level as evidenced by retail prices responding more quickly and more fully to FOB price increases than to FOB price decreases. 30 Li (2007) analyzed the price transmission process for avocados for increases and decreases in shipping- point prices. She summarized the results of her extensive analysis of asymmetric price adjustments for California avocados as follows: The [retail] price adjustment rates were 76% to an increase in shipping price and 29% to a decrease in shipping price, and the adjustment was made slower in response to an increase in shipping price than to a decrease in shipping price. Asymmetry in price adjustment to changes in shipping price suggests that retail- ers were able to manipulate price adjustments to increases and decreases in shipping price to attain higher profits, (p. 333) Thus, retail prices for avocados respond more fully to shipping-point price increases than to shipping- point price decreases. As a result, retail price margins for avocados tend to increase with larger and more frequent price changes or decrease with smaller and less frequent price changes. Price instability promotes higher retailer margins at the expense of both produc- ers and consumers while increased price stability tends to decrease annual average retailer margins with benefits flowing to both producers and consumers. Thus, information programs that smooth the flow of avocados to U.S. markets and the price of those avo- cados benefit both producers and consumers. 7.2. Costs of the Hass Avocado Board's Information Program The annual costs of HAB's information program are listed by category in each HAB annual report (2003-2007) and are summarized in Table 14. Annual expenditures for the information program ranged from $340,179 to $1,090,228 over five years with an average annual cost of just under $750,000. Total five- year costs for the categories of information, analysis, and the Network Marketing Center fell in a rather tight range of $530,514 to $536,936. Almost 57% of the total cost for the first five years ($2,125,915) was in the interaction category. Table 1 4. Annual and Total Costs of the Hass Avocado Board's Information Programs by Cost Category, 2003-2007 Year Cost Category 2003 2004 2005 2006 2007 Grand Total Information $28,619 $219,553 $71,104 $123,434 $94,226 $536,936 Analysis SO $44,843 $168,976 $197,375 $120,281 $531,475 Interaction $286,560 $658,956 $378,566 $404,241 $397,592 $2,125,915 Network marketing center SO $166,876 $66,163 $179,052 $118,423 $530,514 Total information $340,179 $1,090,228 $684,809 $904,102 $730,522 $3,749,840 Source: Hass Avocado Board annual reports, 2003-2007. Studies that have found asymmetry in price transmission for food products include Kinnucan and Forker (1987) for dairy products; Pick, Karrenbrock, and Carman (1990) for citrus; Zhang, Fletcher and Carley (1995) for peanuts; and Carman and Sexton (2005) for fluid milk in the western United States. 34 An Economic Evaluation of the Hass Avocado Promotion Order's First Five Years 7.3. Estimated Benefits from the Information Program We used the results from Li's research on price transmission in the marketing channel to estimate weekly changes in gross marketing margins between the shipping-point price (FOB) and the retail price of avocados. Thus, we assumed that the retail price increased, on average, by 76% of the increase in the shipping-point price and decreased by 29% of the decrease in the shipping-point price. We used the aggregate estimated adjustment without attempting to account for the two to three weeks required for the total price adjustment according to Li's analysis. The changes in estimated gross marketing margins from week to week were based on total weekly shipments, the change in the average weighted shipping-point price per pound for all Hass avocados, and Li's estimated adjustment ratios. Table 15 shows annual estimated gross changes in marketing margins based on each marketing year's weekly total shipments and weighted weekly average shipping-point prices for Hass avocados. The actual annual standard deviations of weekly Hass avocado shipping-point prices both decrease and increase from year to year, ranging from a high of 0.271 in 2003 (the hrst year of the information program) to a low of 0.058 in 2006 (a year of record weekly shipments due to a very large California crop). Estimated total changes in marketing margins associated with changes in the shipping-point price vary from $2,889,059 in 2004 to a little more than $10 million in 2007. Note that total changes in marketing margins are positively related to average weekly shipments and the standard deviation of weekly prices during the marketing year. The standard deviation of weekly prices reported in Table 15 measures actual price variability but we also require an estimate of how different this variability would have been without HAB's information program. In other words, has price variability been reduced by the HAB information program and, if so, by how much? Our approach was to compare the variability of prices immediately before and after initiation of the information program. A limitation of this approach is that the entire change in price variability is attributed to the information program even if there were other factors that contributed to it. As noted, the standard deviation of annual Cali- fornia Hass avocado prices decreased from an annual average of 0.2843 for 1998-2002 to an annual average of 0.2045 for 2003-2007. This decrease of 28% in price variability is used as the maximum reduction in price variability due to HAB's information program. The estimated total increase in marketing margins for 2003-2007 as a consequence of price variability is $31,661,000 (from Table 15). Considering that this hgure represents the reduced value from the presence of the information program, the decrease of 28% in margins would have been worth a five-year (undiscounted) total of $12.3 million in terms of reduced margin that is reflected in both lower retail consumer prices and higher prices to growers at the shipping point. 31 Table 1 5. Estimated Total Annual Changes in Gross Margins for Hass Avocados, Average Shipments, Standard Deviations of Price, and Average Prices for 2003-2007 Year Item Estimate 2003 2004 2005 2006 2007 Margin change (S) 6,533,780 2,889,059 8,133,135 4,033,952 10.070,172 Average weekly shipments (pounds) 8,512,807 11,771,751 12,484,837 15,194,896 13,361,154 Standard deviation of price ($/pound) 0.271 0.128 0.216 0.058 0.263 Average weighted price ($/pound) 1.136 1.018 0.955 0.761 0.993 31 Let M 0 denote the increase in margin due to price variability in the absence of the HAB program and M, = 31,661,000 equal the value 'in the presence of the program. We then have (M 0 - M,) / M„ = 0.28. Solving for M, and subtracting M Q from it yields S12.3 million. 35 Giannini Foundation Research Report 351 The division of the total benefit, as well as the assessment cost to fund the information program, between consumers and producers depends on the value of consumers' price elasticity of demand, e D , relative to producers' price elasticity of supply, e s , for avocados to the U.S. market. As noted in section 5, we have good estimates of e D from the econometric analysis in section 4 but lack a reliable method by which to estimate s s in the current market environ- ment. Thus, section 5 reported benefit-cost ratios for alternative values of s s of 0.5, 1.0, and 2.0. The share of a change in margin going to consumers in terms of lower price is AP= ^ . For purposes of this calculation, we computed e d at the average of price and quantity for the past ten years. Although the estimates vary depending on the specific econometric model estimated, all produced a value of E D ~ -0.25 during this period. Thus, the producers' share of the benefit and the cost from the information program varies from about 11% to 33% depending on the value assumed for e s . Assuming that the entire margin reduction can be attributed to HAB's informa- tion program, the total net benefit is the $12.3 million gross benefit minus the $3.75 million program cost, resulting in $8.55 million of net benefit. Producers' share of this net benefit is then in the range of $0.94 to $2.82 million dollars with the remainder of the net benefit going to U.S. avocado consumers. J6 An Economic Evaluation of the Hass Avocado Promotion Order's First Five Years 8. CONCLUSIONS This study has focused on the impact of HAB promotion programs on demand for avocados. The estimated elasticity of promotion ranged from a high of 0.37 to a low of 0.15 depending on model specification. It appears that the trend variables accounted not only for the effects of "tastes and prefer- ences" but also for some of the effects of promotion. Thus, the low estimate of the promotion elasticity is undoubtedly too low and is viewed as conservative. Simulations of benefit-cost ratios using the highest and lowest estimated promotion response and supply elasticities of 0.5, 1.0, and 2.0 indicate that producer- funded promotion programs not only expanded demand for avocados but provided a positive return on funds spent. The estimated benefit-cost ratios range from 1.12 to 6.73. More importantly, each exceeds 1.0, meaning that (1) the promotion programs supported by HAB during its first five years yielded net benefits to producers and (2) the programs could have been profitably expanded during the 2003-2007 period of analysis. Given the range of promotion and supply elasticities used for the simulations, our best estimate of the benefit-cost ratio for HAB promotion programs is somewhere in the middle of the simulated range of 1.12 to 6.73— most likely, in an interval between 2.5 and 4.0. The orderly marketing objective of HAB's informa- tion program implies a smooth matching of weekly supply and demand with stable prices. Both produc- ers and consumers benefit from price stability when retail prices respond more to FOB price increases than to price decreases as occurs with avocados. A comparison of weekly avocado prices for the five years preceding the creation of HAB with the first five years of its operation shows that price variability decreased an average of 28%. Estimated total producer and consumer benefits from HAB's information program may have been as much as $12.3 million. Subtracting $3.75 million for program costs leaves a net benefit of $8.55 million. Producers' share of this net benefit is estimated to be in the range of $0.94 to $2.82 million with the remainder of the net benefit going to U.S. avocado consumers. Analysis of avocado promotion programs in major retail markets suggests that radio promotion significantly increased average retail sales in promo- tion markets over the same sales in nonpromotion markets. Previous results also suggested that radio is a more effective medium than outdoor advertising but the difference in effect was not statistically significant. The opportunity to conduct an evaluation based on available retail scanner data was limited by the indus- try's inability to systematically provide disaggregate promotion expenditure information. A potential problem with producer-funded con- sumer promotion programs is that retailers may respond to increased demand by raising retail prices, thereby curtailing the demand expansion. Analysis of retail pricing behavior for avocados indicates, however, that this does not seem to have occurred sys- tematically and, thus, was not a problem limiting the impact of promotion. Instead, there is weak evidence that retailers may contribute to the effectiveness of the programs by lowering prices to support industry promotions. 37 Giannini Foundation Research Report 351 An Economic Evaluation of the Hass Avocado Promotion Order's First Five Years REFERENCES Ackerberg, D.A. "Empirically Distinguishing Informa- tive and Prestige Effects of Advertising." RAND Journal of Economics 32 (2001):316-333. Ackerberg, D.A. "Advertising, Learning, and Con- sumer Choice in Experience Good Markets: An Empirical Appraisal." International Economic Review 44 (2003): 1007- 1040. Alston, J.M., J.A. Chalfant, J.E. Christian, E. Meng, and N.E. Piggott. The California Table Grape Commission's Promotion Program: An Evaluation. Davis CA: Giannini Foundation of Agricultural Economics Monograph 43, 1997. Alston, J. M., H.F. Carman, J.A. Chalfant, J. M. Crespi, R.J. Sexton, and R.J. Venner. The California Prune Board's Promotion Program: An Evaluation. Davis CA: Giannini Foundation of Agricultural Economics Research Report 344, 1998. AlstonJ.M.J.M. Crespi, H.M. Kaiser, and R.J. Sexton. "The Economics of Commodity Promotion Pro- grams: Summary and Synthesis." The Economics of Commodity Promotion Programs: Lessons from California. H.M. Kaiser et al, eds., pp. 387-415. New York NY: Peter Lang Publishing, Inc., 2005. California Avocado Commission. "Annual Report." Irvine CA, 2003/04-2007/08. Available at www. avocado.org/industry/ commission/ annual- reports. California Avocado Commission. "Business Plan." Irvine CA, 2004/05-2006/07. Available at www.avocado.org/industry/commission/annual- reports. Carman, H.F. "Offsetting Price Impacts from Imports with Generic Advertising and Promotion Programs: The Hass Avocado Promotion and Research Order." Review of Agricultural Economics 28(4) (2006):463-481. Carman, H.F, and R. Cook. "An Assessment of Potential Economic Impacts of Mexican Avocado Imports on the California Industry." Proceedings of the ISHS Xlllth International Symposium on Horticultural Economics. Leiden, The Netherlands: Interna- tional Society for Horticultural Science, 1996, pp. 227-234. Carman, H.F., and R.K. Craft. An Economic Evaluation of California Avocado Industry Marketing Programs, 1961-1995. Davis CA: Giannini Foundation of Agricultural Economics Research Report 345, 1998. Carman, H.F, and R.D. Green. "Commodity Supply Response to a Producer Financed Advertising Program: The California Avocado Industry." Agribusiness 9 (1993):605-621. Carman, H.F, and R.J. Sexton. "Supermarket Fluid Milk Pricing Practices in the Western United States." Agribusiness 21(4) (2005):509-530. Chilean Avocado Importers' Association. "Annual Financial Statement." Washington DC, 2003- 2007. Cook, R. "The Avocado Market: A Growth Market in a Mature Food Industry." Avocado Brainstorming Session, University California, Davis, Department of Agricultural and Resource Economics, 2003. Available at http://ucce.ucdavis.edu/hles/data- store/234-65.ppt. Crespi, J. M., and R.J. Sexton. "Evaluating the Effective- ness of California Almond Promotion: How Much Did the Litigation Cost Producers?" The Economics of Commodity Promotion Programs: Tessons from California. H.M. Kaiser, et al., eds., pp. 167-190. New York NY: Peter Lang Publishing, Inc., 2005. Erdem, T, and M.P Keane. "Decision-making under Uncertainty: Capturing Dynamic Brand Choice Processes in Turbulent Consumer Goods Mar- kets." Marketing Science 15 (1996): 1-20. Hass Avocado Board. "Annual Report." Irvine CA, 2003-2007. Hass Avocado Board. "Domestic Marketing Plan." Irvine CA, 2004/05-2006/07. Kaiser, H.M., J.M. Alston, J.M. Crespi, and R.J. Sex- ton, eds. The Economics of Commodity Promotion Programs: Tessons from California. New York NY: Peter Lang Publishing, Inc., 2005. Kinnucan, H.W., and O. Forker. "Asymmetry in Farm Retail Price Transmission for Major Dairy Prod- ucts." American Journal of Agricultural Economics 69 (1987):285-292. 39 Giannini Foundation Research Report 351 Kinnucan, H.W., and Y. Zheng. "National Benefit-cost Estimates for the Dairy, Beef, Pork, and Cotton Promotion Programs." The Economics of Commod- ity Promotion Programs: Lessons from California. H.M. Kaiser, et al., eds., 255-284. New York NY: Peter Lang Publishing, Inc., 2005. Li, L. "Retailer Pricing Behavior for a Fresh Produce Commodity: The Case of Avocados." Ph.D. disser- tation, University of California, Davis, 2007. Mexican Hass Avocado Importers' Association. "Hass Avocado Marketing Campaign Plan." Prepared by the Preston/Tully Group, Inc., Fallston MD, 2003-2007. Mexican Hass Avocado Importers' Association. "Letter of Agreement and Budget Allocation between MHAIA and the Preston/Tully Group, Inc." Fall- ston MD, 2003-2007. Pick, D., J. Karrenbrock, and H. Carman. "Price Asymmetry and Marketing Margin Behavior: An Example for California-Arizona Citrus." Agribusiness 6 (1990): 75-84. Sexton, R.J., and M. Zhang. "A Model of Price Deter- mination for Fresh Produce with Application to California Iceberg Lettuce." American Journal of Agricultural Economics 78(4) (1996):924-934. U.S. Department of Agriculture. Regulatory Impact and Regulatory Flexibility Analyses: The Potential Economic Impact of Expanded Importation of Hass Avocados from Mexico. Washington DC: Animal and Plant Health Inspection Service, 2001. U.S. Department of Agriculture. "Proposed Rule to Allow Fresh Hass Avocados Grown in Approved Orchards in Approved Municipalities in Michoa- can, Mexico, to Be Imported into All States Year-Round." Animal and Plant Health Inspection Service, APHIS Docket no. 03-022-3, Washington DC, May 19, 2004. U.S. Department of Agriculture. Animal and Plant Health Inspection Service. Importation of "Hass" Avocado Fruit (Persea Americana) from Mexico: A Risk Analysis. Washington DC: Animal and Plant Health Inspection Service, Plant Protection and Quarantine, 2003. U.S. Department of Agriculture, Foreign Agricultural Service. "World Avocado Situation and Out- look" World Horticultural Trade & U.S. Export Opportunities. Washington DC, 2005. Wolfe, K., and C. Ferland. "Estimated Fresh Salsa Mar- ket Potential for Georgia and Bordering States." University of Georgia Department of Agricultural and Applied Economics report FR-03-01, 2003. Available at www.agecon.uga.edu/~caed/Salsa Industry.pdf. Zhang, P., S. Fletcher, and D. Carley. "Peanut Price Transmission Asymmetry in Peanut Butter." Agribusiness 11(1) (1995):13-20. 40 Giannini Foundation Research Report 351 GIANNINI FOUNDATION RESEARCH REPORT SERIES The Giannini Foundation Research Report Series (ISSN 0899-3068) is designed to communicate research to specific professional audiences interested in applications. The first Research Report was issued in 1961 as No. 246, continuing the numbering of the Giannini Foundation Mimeographed Report Series, which the Research Report replaced. The Giannini Foundation of Agricultural Economics was founded in 1930 from a grant made by the Bancitaly Corporation to the University of California in tribute to its organizer and past president, Amadeo Peter Giannini of San Francisco. The broad mission of the foundation is to promote and support research and outreach activities in agricultural economics and rural development relevant to California. In line with those goals, the foundation encourages research in various areas of interest to agricultural and resource economists and supports dissemination of research findings to other researchers and to the public. Foundation membership includes agricultural economists (faculty and Cooperative Extension specialists) at the Department of Agricultural and Resource Economics, Davis, and at the Department of Agricultural and Resource Economics, Berkeley. Associate members include forestry economists in the College of Natural Resources, Berkeley, and economists in the Department of Environmental Sciences at Riverside. This and other Giannini Foundation publications are available in PDF format online at http://giannini.ucop.edu/publications.htm. Hard copies of this report maybe ordered from University of California Agriculture and Natural Resources Communication Services. University of California Agriculture and Natural Resources (ANR) Communication Services 6701 San Pablo Avenue, 2nd Floor Oakland, California 94608-1239 Telephone: 800.994.8849 / 510.642.2431 Fax: 510.643.5470 E-mail: danrcs@ucdavis.edu Visit the ANR Communication Services Web site at http://anrcatahg.ucdavis.edu Peter Berck Giannini Foundation Series Editor University of California, Berkeley Julian Alston Associate Editor University of California, Davis Julie McNamara Managing Editor University of California, Davis Angie Erickson Technical Editor University of California, Berkeley Natalie Karst Production Visit the Giannini Foundation Web site at http://giannini.ucop.edu UNIVERSITY OF CALIFORNIA, DAVIS 175 03322 7862 The University of California prohibits discrimination or harassment of any person on the basis of race, color, national origin, religion, sex, gender identity, pregnancy (including childbirth, and medical conditions related to pregnancy or childbirth), physical or mental disability, medical condition (cancer-related or genetic characteristics), ancestry, marital status, age, sexual orientation, citizenship, or service in the uniformed services (as defined by the Uniformed Services Employment and Reemployment Rights Act of 1994: service in the uniformed services includes membership, application for membership, performance of sendee, application for service, or obligation for service in the uniformed services) in any of its programs or activities. University policy also prohibits reprisal or retaliation against any person in any of its programs or activities for mak- ing a complaint of discrimination or sexual harassment or for using or participating in the investigation or resolution process of any such complaint. University policy is intended to be consistent with the provisions of applicable State and Federal laws. Inquiries regarding the University's nondiscrimination policies may be directed to the Affirmative Action/Equal Oppor- tunity Director, University of California, Agriculture and Natural Resources, 1111 Franklin Street, 6th Floor, Oakland, CA 94607, (510) 987-0096. For information on obtaining copies of this publication, see the back cover.