Printer Friendly

Impact of hurricanes on frozen concentrated orange juice future prices return.


Oranges are cultivated all over in the tropical and sub-tropical regions of the world. The states of Sao Paulo (Brazil) and Florida (U.S.) are the two leading orange producing areas in the world, and account for over 80 per cent of the world orange juice production. Mexico's production follows these two countries. Spain, China, Italy, and Egypt are the other main orange producing countries in the world (Muraro, 2003; Spreen, 1996). It has been found that Brazil and Mexico are the major exporters of orange juice to the U.S. (Davis A., 2008). And, the oranges produced in Brazil and US (in the state of Florida) are essentially used for processing (Davis A., 2008 and Spreen T., 1996).

Orange juice can be categorized into three main types: not-from-concentrate orange juice (NFC), frozen concentrated orange juice (FCOJ), and orange juice from concentrate (RECON) that needs to be refrigerated. Almost all the orange juice bought in the US belongs to one of these types (Davis A., 2008). To help produce mass quantities of Frozen concentrated orange juice (FCOJ), the process of heat packaging was developed (Ahles A., 2009).

FCOJ future market is a weather-sensitive market and the FCOJ prices are highly volatile. Ninety-nine percent of the FCOJ production is taking place in Florida, particularly in the Orlando region. This geographical concentration is unusual for an agricultural commodity, thus providing an impetus to study the relationship between the asset return, which is the FCOJ future prices, and the weather which serves as the exogenous variable (Ref).

The first research work that focused on how the weather affects the FCOJ return volatility was done by Roll study (1984). The study shows that "the small predictive power of temperature and rainfall implies that there are factors other than the weather that affect FCOJ returns". Roll also reports the results of a regression analysis with the FCOJ return as the dependant variable, and variables such as oil prices return, stock return and Canadian demand as the independent variables.

Roll finds the FCOJ futures market puzzling, because, although weather has the most significant influence on orange crop production, it explains only a small fraction of the observed variability in futures prices. Following the Roll's (1984) methodology, this study investigates the relationship between hurricanes (predicted and observed) with the volatility of the FCOJ prices return.


Major suppliers of orange are Florida State, in the U.S., and Sao Paulo State, in Brazil. The two countries are the two leading orange producing areas in the world (Muraro, R., 2003). The oranges produced here are essentially used for processing (Spreen, T., 1996). Most of the FCOJ produced in Florida is consumed within the U.S. On the other hand, Brazil exports most of its production, accounting for eight percent (80%) of FCOJ produced (Muraro, R., 2003).

According to the Agricultural Crop Forecast Production (Bauer & Orazem 1994)), Florida's citrus-bearing acreage is down from last season's by 65,000 acre. This is the lowest bearing acreage recorded since the 1993-94 season as presented in Appendix 1. It takes between three to five years for newly planted trees to bear fruits of commercially harvestable quantity and fifteen to twenty years to reach their peak production. Orange production requires an extended development time and long-term commitments of land and labor, the factors affecting the supply side. U.S. oranges utilized in production id also depicted in Appendix 2. Therefore, such news about orange corps may move the price. Bauer and Orazem (1994) concluded that there is a strong negative correlation between production level surprises and FCOJ futures returns.

Brazil supplies thirty-fifty percent (30-50%) of the FCOJ consumed in the U.S. Since supply from Brazil happens to be insensitive to any weather conditions, the impact of weather is likely to be less on supply levels and consequently prices (Fleming, J., 2006). Since May 2005, the primary FCOJ contract traded in the New York Board of Trades (NYBOT) specifies that juice deliverable against the contract must be of Florida or Brazil origin. Similar to freeze season in Florida mainly in December, January, and February (Fleming, J., 2006) there is a drought season in Brazil that lasts from July through November.

The Brazilian orange industry has a cost advantage in sending bulk FCOJ to the U.S. as well as the European markets. The present US tariffs on FCOJ has allowed Florida's orange juice industry to compete with Brazil (for FCOJ shipped to the northeastern US market) as well as in the European market, and thus helped it in its profitability (Muraro, R., 2003).


First, the study identifies the vis-a-vis, the market price. The study then identifies the factors that generate a shock to the supply and demand dynamics. Finally, it determines whether those factors affect the return volatility of the FCOJ or not. If so, is the volatility a result of investors' irrational behavior or could it be due to market-microstructure reason.

Weather--The important fundamental factor for the FCOJ futures market is the weather shock on production of orange crops. Roll (1984) provided evidence that there is a statistically significant relation between errors in temperature and orange production forecasts issued by the National Weather Service for the central Florida region. He further concluded that there is no significant statistical association between orange juice prices and rainfall predictions.

In weather-sensitive markets, weather conditions in a concentrated geographical region can have a great effect on the supply and/or demand and hence price. In weather sensitive FCOJ market, trading versus non-trading ratio is lower than that in equity market but higher than that in the currency market. The variance ratio is found to be considerably lower at that time of the year when prices are extremely sensitive to weather. These facts confirm that there is a strong relationship between prices of FCOJ and public information availability, and that seasonality in the variance ratio cannot be related to pricing errors or seasonal variations in trading levels (Fleming J., 2006).

Temperature--Freeze can damage or destroy growing fruits on trees or can destroy the bloom and the new growth, thus reducing expected production in the following season. Fruit damage means a decreasing juice content of the surviving fruits and if the freeze is severe, large portion of the fruits may not be usable at all. Freeze season in Florida runs from December through March while the US marketing season for FCOJ begins December 1st and continues through November 30th. Oranges in Florida are harvested mainly from November through June (Ahles A., 2009).

Spreen T., Brown M., and Lee J. (1996) in their paper have stated that a series of freezing temperatures affected the orange producing areas of Florida in 1977, 1981, 1982, 1983, 1985, and 1989. As a result, not only the production of oranges declined, but millions of orange trees were destroyed. In short, the orange producing ability of the state of Florida was reduced. Furthermore, the reduced supply resulted in constantly high prices of orange juice throughout the latter half of the 1980s. In turn, these elevated price-levels led to a quick expansion of orange production levels in Brazil (Spreen T., 1996).

Table 1 shows how below freezing temperatures affect orange production.

Hurricanes--Florida is a major target for hurricanes, leading to the study's hypothesis that the FCOJ futures market is very sensitive to the hurricanes season predicted and observed. FCOJ prices rose by over 50 percent during the year ending in June 2005, because of a fear of hurricanes.

Disease: Like citrus production cranker has a considerable economic impact on the annual acre-return. According to the department of Food and Economic Resources at Florida University, it is estimated that the total production of early mid-varieties could decrease by 10 percent and the total production of the Valencia oranges could decrease by 5 percent due to this phenomenon.

U.S. Economy--The strength of the US economy and the US dollar affect the supply and pricing of FCOJ, as FCOJ is priced for international trade in US dollars.

Demand Factors--Consumer preference can affect pricing substantially. But since change in consumer preference and diet behavior takes a long time to change it is assumed this factor does not constitute a demand shock. Also, change in demographics has been found to be related to change in the demand for orange juice. When the percentage of Asians increase in a city, the demand for orange juice rises, but when the percentage of African American and Hispanic population rises, the demand for orange juice declines (Davis, A., 2008).



Future contracts in FCOJ have been traded on the NYBOT since 1967. Usually at any given time there are nine contracts with expiration (delivery scheduled) every second month i.e. January, March, May, July, September, November, with at least two contracts listed for the month of January at all times. A contract is usually for 15,000 pounds (3 percent more or less) of solid concentrated orange juice (all water being removed) with standard requirements for grade, Brix value and color. There are two types of oranges produced in that region, early and midseason varieties which are harvested from November to March, and Valencia oranges which are harvested from April to June.

The daily closing prices of FCOJ were obtained for the period October 1, 2002 to May 1, 2005 are presented in Figure 1 and Figure 2. These were actually five closing prices of the near five contracts from New York Board of Trade NYBOT. For June 1, 2002, for example, we accessed the closing prices for July 2002 contract, September 2002 contract, November 2002 contract, January 2003 contract and March 2003 contract, respectively.


By examining the data, available (the volume and open interest) for the purpose of calculating the return, only the second nearby, third nearby and the fourth nearby contracts were chosen. The fact that the first nearby contract has the highest liquidity since it is close to maturity and the fifth contract does not have much activity since it has less volume and less open interest, being far from maturity were ignored.



The temperature data was obtained from the The data consists of the daily maximum and the minimum temperatures as presented in Figure 3.


The data for hurricanes was obtained from the hurricane center at Colorado State University. The data consists of the number of intense hurricanes predicted, the day prediction was announced and the hurricanes actually observed each year, between 2000 and 2006, for the Gulf Coast and the Florida regions, as presented in Table 2.

Further data for the observed hurricanes for the Orlando region was obtained from the

National Oceanic and Atmospheric Administration. (Using ho indicates observed hurricanes, while hp represents predicted hurricanes.)

Production Forecast

The year-end production forecast data and the final actual production data were obtained from the National Agriculture Statistics Service which is a part of the USDA. The USDA announced the year-end forecasts on the first day of every month, from October through July. For example, the first forecast for the year 2004-2005 production was announced on October 1st, 2004. Thus we have nine forecast announcements for each production year. The production forecast shock: Production = Monthly Forecast--Final Production, for the announcement day, and zero for the other days as shown in the graph of Figure 4.




Regression analysis (using ordinary least square) is applied to determine if the prediction of hurricanes really affect the future market for the FCOJ. Significant results are not expected because of inadequacy or absence of data. Nevertheless, the study should provide an impetus for further research in this important area.

Starting with a simple equation to regress return on temperature.

[R.sup.2.sub.t] = [alpha] + [beta][temp.sub.t-1] + [[epsilon].sub.t]

Return for a contract is given by, [R.sub.i] = ([p.sub.i,t] - [p.sub.i,t-1])/[p.sub.i,t-1]

Here i represent contracts 2, 3 and 4. These are appropriate to close returns.

The average return is, [R.sup.*] = ([4.summation over (i=2)] [R.sub.i])/3

Temperature is calculated as, temp = max (0,32 _ Minimum)

Thus, we get a zero for all temperature data series except for the days when the temperature is at or below freezing level i.e. 32 degrees Fahrenheit.

FCOJ returns and the temperature (aligned in time) show a very low R-squared value (0.000025) indicating that the temperature alone does not explain much of the volatility in the return even when the temperatures used are actual temperatures recorded (and not the predicted ones), Durban-Watson statistics was about 1.75, as presented in Table 3.

Dependent Variable: [R.sup.2.sub.t] Sample(adjusted): 10/02/2002 4/29/2005

Included observations: 673 after adjusting endpoints

In the second equation, we add the production shock to the first equation.

[R.sup.2.sub.t] = [alpha] + [[beta].sub.1][temp.sub.t-1] + [[beta].sub.2] [production.sub.t-1] + [[epsilon].sub.t]

Adding the production variable in the first regression equation does not explain any better the volatility in return, as the value of R squared increased slightly from 0.000025to 0.000135. Durban-Watson value remained around 1.74 as presented in Table 4.

Finally, we added to the second equation the important variables of predicted and observed hurricanes in two separate regressions respectively.

[R.sup.2.sub.t] = [alpha] + [[beta].sub.1][temp.sub.t-1] + [[beta].sub.2][production.sub.t-1] [[beta].sub.3][predictive.sub.t-1] + [[epsilon].sub.t]

Dependent Variable: [R.sup.2.sub.t] Sample(adjusted): 10/02/2002 4/29/2005 Included observations: 673 after adjusting endpoints

[R.sup.2.sub.t] = [alpha] + [[beta].sub.1][temp.sub.t-1] + [[beta].sub.2][production.sub.t-1] + [[beta].sub.3][observed.sub.t-1] + [[epsilon].sub.t]

Dependent Variable: [R.sup.2.sub.t] Sample(adjusted): 10/02/2002 4/29/2005 Included observations: 673 after adjusting endpoints

Results of adding the hurricane variables indicate that the FCOJ return relationships is not highly significant. However, it is worth noting that the R-Squared value obtained when using actual hurricane variable (data) is significantly higher when compared with the R-Squared value obtained using the predicted hurricane variable, which appears in Table 5 and Table 6.

Table 7 summarizes the results of the above four regression analyses.


The decrease in the acreage-bearing orange trees is one of the endogenous variables which affect the orange crops while keeping in mind the fact that the citrus disease (cranker) destroys the crops and trees and put limitations on planting new orange trees. This explains why there was a dramatic decrease in the crop during this season as well as in the previous one. The FCOJ return is affected by the freeze temperatures. But the temperature does not explain much of the volatility in FCOJ returns. Further the FCOJ return volatility has significant relationship to the variation between the forecasted production, which is announced in the beginning of each month, and the end-of-season production.

Predictive hurricanes are statistically significant but do not add to the explanation of volatility in the FCOJ return. Observed hurricanes, on the other hand, do increase the explanation of the return volatility of FCOJ. As Roll concluded even though the weather has the most significant influence on the orange crop production, it still remains a puzzle in the analysis of FCOJ futures market.


U.S. Oranges Utilized Production

Year     Tons

 96     11,426
 97     12,692
 98     13,670
 99      9,842
 00     12,997
 01     12,221
 02     12,374
 03     11,545
 04     12,872
 05      9,252
 06      8,895


Ahles, A. (2009). The Globalization of Orange Juice, Sociology: Sociology & the Politics of Food, 482-582.

Bauer, R., & Orazem, P. (1994). The rationality and price effects of U.S. Department of Agriculture forecasts of oranges, Journal of Finance, 49, 681-695.

Beck, S. (1994). Cointegration and market efficiency in commodities futures markets, Applied Economics, 23, 249-257.

Boudoukh, R., Shen, & Whitelaw (2002). Testing the Market Rationality: Lessons from the FCOJ Market, Stern school of business (working paper).

Boudoukh, Richardson, Shen, & Whitelaw (2002). Do Asset Prices Reflect Fundamentals? Freshly Squeezed Evidence from the FCOJ Market, Stern School of Business.

Davis, A., Gunderson, M., Brown, M., & House, L. (2008). The effect demographics have on the demand for orange juice, Paper presented at the Southern Agricultural Economics Association Annual Meeting, Dallas, Texas, February 2-6, 2008.

Fleming, J., Kirby, C., & Ostdier, B. (2006). Information, trading and volatility: evidence from weather-sensitive markets, The Journal of Finance, LXI (6), 2899-2918.

Muraro, R., & Spreen, T., (2003). Comparative marketing costs for FCOJ from Florida and Sao Paulo, EDIS Document FE363 publication of the Department of Food and Resource Economics, Institute of Food and Agricultural Sciences, University of Florida, Gainsville, Florida.

Roll, R. (1984). Orange Juice and Weather, American Economic Review, 74, 861-880.

Spreen, T., Brown, M., & Lee, J. (1996). The Impact of NAFTA on U.S. Imports of Mexican Orange Juice. Retrieved on March 01, 2011.

Moussa Fouad is an Assistant Professor of Finance, Southern University at New Orleans, Louisiana.

Waheed Abdul is an Associate Professor in the College of Business, Southern University at New Orleans, Louisiana.

Fouad Moussa

Abdul Waheed

Southern University at New Orleans
Table 1
Critical Temperature for Oranges

Temperature        Consequences

28            Extensive Fruit Damage
26            Extensive Fruit Damage
24            Extensive Fruit Damage
20            Extensive tree Damage

Table 2 Intense Hurricanes

year   Prediction      Date      Observed       Hurricanes and
          (hp)      prediction     (ho)         tropical storms
                    announced                    Observed for
                                                Orlando region
2000   4 3          August 4        3
2001   3 3          August 7        4       Sept 14,15 Gabrielle TS
2002   1 1          August 7        2       Sept 4,5 Edward TS
                    Sept 2
2003   3 3 2        August 6        3
                    Sept 3
                    Oct 2
2004   3 5 6        August 6        6       August 13,14 Charley H
                    Sept 3                  Sept 5 Francis H
                    Oct 1                   Sept 26 Jeanne H

2005   6 6 6        August 5        7       October 5 Tammy H
                    Sept 2
                    Oct 3
2006   3            August 1     Not available

Table 3 Regression Result with Temperature Variable

Variable         Coefficient   Std. Error   t-Statistic    Prob.

C                 0.000262     2.38E-05      11.00279       0.0000
TEMP(-1)          1.59E-05     0.000123      0.128582       0.8977
R-squared         0.000025     Mean dependent var           0.000262
Adjusted         -0.001466     S.D. dependent var           0.000616
S.E. of           0.000616     Akaike info criterion      -11.94271
Sum squared       0.000255     Schwarz criterion          -11.92930
Log likelihood    4020.721     F-statistic                  0.016533
Durbin-Watson     1.745568     Prob(F-statistic)            0.897727

Table 4
Regression Result with Temperature and Production Variables

Variable         Coefficient   Std. Error   t-Statistic   Prob.

C                0.000262      2.38E-05     10.98789      0.0000
TEMP(-1)         1.58E-05      0.000123      0.127769     0.8984
PRODUCTION(-1)  -3.43E-06      1.26E-05     -0.272460     0.7854
R-squared        0.000135      Mean dependent var         0.000262
Adjusted        -0.002849      S.D. dependent var         0.000616
S.E. of          0.000617      Akaike info criterion    -11.93985
Sum squared      0.000255      Schwarz criterion        -11.91973
Log likelihood   4020.758      F-statistic                0.045373
Durbin-Watson    1.744492      Prob(F-statistic)          0.955644

Table 5 Regression Result with Temperature, Production
and Predictive Hurricane Variables

Variable         Coefficient   Std.Error   t-Statistic   Prob.

C                0.000250      2.72E-05    9.176928      0.0000
TEMP(-1)         1.82E-05      0.000123    0.147558      0.8827
PRODUCTION(-1)  -4.70E-06      1.26E-05   -0.372019      0.7100
PREDICTIVE(-1)   1.25E-05      1.34E-05    0.933303      0.3510
R-squared        0.001436      Mean dependent var        0.000262
Adjusted        -0.003042      S.D. dependent var        0.000616
S.E. of          0.000617      Akaike info criterion    -11.93818
Sum squared      0.000255      Schwarz criterion        -11.91136
Log likelihood   4021.196      F-statistic               0.320594
Durbin-Watson    1.747053      Prob(F-statistic)         0.810492

Table 6 Regression Result with Temperature,
Production and Observed Hurricane Variables

Variable         Coefficient   Std. Error    t-Statistic   Prob.

C                0.000253      2.35E-05      10.76379      0.0000
TEMP(-1)         1.77E-05      0.000121       0.145870     0.8841
PRODUCTION(-1)  -3.08E-06      1.23E-05      -0.249905     0.8027
OSERVED(-1)      0.001582      0.000303       5.214552     0.0000
R-squared        0.039188      Mean dependent var          0.000262
Adjusted         0.034879      S.D. dependent var          0.000616
S.E. of          0.000605      Akaike info criterion      -11.97672
Sum squared      0.000245      Schwarz criterion          -11.94990
Log likelihood   4034.165      F-statistic                 9.095280
Durbin-Watson    1.905200      Prob(F-statistic)           0.000007

Table 7
Regression Analysis Summary

        Temp     Production   PREDICTIVE   OBSERVED   Adjusted
        (-1)        (-1)         (-1)        (-1)     R-squared

Eq1   0.128582                                        -0.001466
Eq2   0.127769   -0.272460                            -0.002849
Eq3   0.147558   -0.372019     0.933303               -0.003042
Eq4   0.145870   -0.249905                 5.214552    0.034879

      F-statistic     Durbin-
                    Watson stat

Eq1   0.016533       1.745568
Eq2   0.045373       1.744492
Eq3   0.320594       1.747053
Eq4   9.095280       1.905200
COPYRIGHT 2011 International Academy of Business and Public Administration Disciplines
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2011 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Moussa, Fouad; Waheed, Abdul
Publication:International Journal of Business and Economics Perspectives (IJBEP)
Article Type:Report
Geographic Code:1USA
Date:Sep 22, 2011
Previous Article:Is saving generating economic growth in Lebanon?
Next Article:Ethical aspects of economic sanctions: a third theory.

Terms of use | Privacy policy | Copyright © 2021 Farlex, Inc. | Feedback | For webmasters