Citation
Measuring the Profitability of the United States Food Supply Chain

Material Information

Title:
Measuring the Profitability of the United States Food Supply Chain Cross-Section and Time-Series Effects
Creator:
Zhao, Youshan
Place of Publication:
[Gainesville, Fla.]
Publisher:
University of Florida
Publication Date:
Language:
english
Physical Description:
1 online resource (74 p.)

Thesis/Dissertation Information

Degree:
Master's ( M.S.)
Degree Grantor:
University of Florida
Degree Disciplines:
Food and Resource Economics
Committee Chair:
Weldon, Richard N.
Committee Members:
Randles, Ronald H.
Trejo, Carlos
Graduation Date:
8/7/2010

Subjects

Subjects / Keywords:
Agribusiness ( jstor )
Assets ( jstor )
Business structures ( jstor )
Economic recessions ( jstor )
Financial ratios ( jstor )
Food ( jstor )
Food supply ( jstor )
Supply chain management ( jstor )
Time series ( jstor )
Wholesale trade ( jstor )
Food and Resource Economics -- Dissertations, Academic -- UF
agribusiness, cross, dupont, financial, panel, profitablility, time, u
Genre:
Electronic Thesis or Dissertation
bibliography ( marcgt )
theses ( marcgt )
government publication (state, provincial, terriorial, dependent) ( marcgt )
Food and Resource Economics thesis, M.S.

Notes

Abstract:
Over time, agricultural producers and agribusinesses in the U.S. food supply chain have experienced a lot of market changes and restructurings. Agribusiness management is increasingly focusing on the strategy to improve its profitability. There are many key drivers to improve agribusiness s competitiveness and financial performance, such as market demand, technical efficiency, environmental impact, etc. This study concentrates on the operating strategy employed by the U.S. food supply sectors by analyzing groups of financial ratios. Following the framework of the DuPont system, this research classifies financial ratios into three categories to measure agribusiness s operating efficiency, asset use efficiency, and capital structure. Moreover, both fixed cross-section effect and time-series effect are examined though the financial statement analysis with respect to a 23 years panel data. This research aims to: 1) evaluate the critical operating concern for profitability of agribusiness; 2) compare the profitability of sub sectors in the food supply chain; and 3) estimate the time effect of economic recessions on agribusiness s profit rate. This study provides information regarding the financial performance of agribusiness. And it advances the understanding of operating efficiency and effectiveness of agribusiness. ( en )
General Note:
In the series University of Florida Digital Collections.
General Note:
Includes vita.
Bibliography:
Includes bibliographical references.
Source of Description:
Description based on online resource; title from PDF title page.
Source of Description:
This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Thesis:
Thesis (M.S.)--University of Florida, 2010.
Local:
Adviser: Weldon, Richard N.
Statement of Responsibility:
by Youshan Zhao.

Record Information

Source Institution:
UFRGP
Rights Management:
Applicable rights reserved.
Embargo Date:
10/8/2010
Resource Identifier:
004979976 ( ALEPH )
707467133 ( OCLC )
Classification:
LD1780 2010 ( lcc )

Downloads

This item has the following downloads:


Full Text







30%

25%

20%

15%

10%

5%

0%
-5%1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010

-10%
Food retailers(54) food wholesalers(514) Food services(581)


Figure 3-3. EBTOCE fluctuation by weighted average in the FWRS industry










Table 3-3. EBTOCE of FPB industry by year and sub sector
SIC 201 202 203 204 205 206 207 208 209 21

C.F, and Sugar and Fats and Miscellane
Year Meat Diary PF and V Grain mill Bakery confectionery oils Beverages ous food Tobacco Mean


1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
MEAN
Std


15.92%
14.63%
11.89%
9.90%
12.81%
7.57%
9.56%
12.04%
17.12%
18.96%
11.79%
8.77%
11.27%
15.14%
10.57%
8.31%
6.12%
7.60%
11.34%
10.60%
1.13%
6.61%
-9.20%
10.02%
5.75%


21.87%
19.76%
24.10%
20.30%
19.40%
16.56%
13.41%
13.36%
12.12%
-4.75%
12.05%
10.29%
5.23%
9.06%
8.14%
4.08%
8.68%
9.90%
7.07%
6.01%
9.73%
7.52%
5.08%
11.26%
6.76%


18.44%
21.20%
20.49%
18.11%
18.04%
23.39%
21.49%
18.83%
20.78%
20.14%
29.46%
19.64%
20.14%
25.05%
21.44%
24.84%
17.60%
19.49%
16.31%
14.24%
16.13%
16.32%
15.22%
19.86%
3.51%


24.97%
29.64%
25.19%
22.04%
24.54%
24.98%
23.87%
21.84%
21.49%
30.72%
21.24%
8.82%
20.05%
24.36%
24.92%
9.51%
11.84%
12.55%
13.78%
10.59%
11.05%
13.56%
12.51%
19.31%
6.83%


15.48%
14.77%
17.73%
18.70%
16.28%
9.09%
9.48%
4.76%
6.51%
7.48%
4.76%
6.74%
3.01%
6.63%
8.96%
6.52%
4.01%
1.21%
13.43%
14.13%
15.46%
17.08%
15.49%
10.34%
5.37%


18.34%
42.88%
24.13%
24.58%
24.78%
21.26%
19.95%
22.43%
21.32%
21.23%
21.90%
28.18%
20.34%
25.71%
19.69%
19.83%
21.61%
15.87%
17.04%
17.12%
16.89%
12.46%
9.20%
21.16%
6.36%


13.41%
15.13%
16.22%
17.13%
16.96%
14.11%
12.67%
10.63%
9.68%
14.48%
11.92%
7.06%
6.03%
3.70%
2.88%
3.67%
7.05%
5.73%
5.91%
11.64%
12.95%
19.50%
13.37%
10.95%
4.84%


14.65%
17.23%
18.76%
17.52%
18.10%
17.98%
18.86%
19.62%
20.76%
19.94%
18.98%
21.66%
18.50%
16.47%
15.21%
18.75%
18.71%
16.32%
17.68%
17.16%
16.81%
18.62%
9.55%
17.73%
2.42%


12.39%
15.19%
11.45%
14.06%
15.30%
16.79%
16.82%
14.60%
9.08%
12.72%
6.89%
13.79%
16.33%
17.64%
24.31%
20.20%
18.85%
18.46%
17.10%
22.75%
12.09%
14.20%
12.66%
15.38%
4.04%


21.30%
24.76%
15.02%
13.84%
15.40%
15.67%
18.69%
13.80%
17.79%
20.13%
23.47%
22.26%
18.18%
21.72%
19.45%
16.92%
19.28%
12.78%
15.84%
16.72%
17.55%
20.06%
31.43%
18.78%
4.20%


17.68%
21.52%
18.50%
17.62%
18.16%
16.74%
16.48%
15.19%
15.66%
16.11%
16.25%
14.72%
13.91%
16.55%
15.56%
13.26%
13.38%
11.99%
13.55%
14.10%
12.98%
14.59%
11.53%
15.48%
2.33%









financial performance for sub-sectors of food supply chain by performing the fixed effect

panel regression.

The third hypothesis takes the following form:

Hypothesis 3

The sub sectors' profitability does not vary across time. There are no uniform time

effects to those sub sectors.

The rest of this paper is organized as follows. Chapter 2 reviews the literature,

introduces methodology and specifies the model. Chapter 3 summarizes the data

selection criteria and provides descriptive statistics. In Chapter 4, hypotheses stated in

Chapter 1 are tested. Empirical results are analyzed in Chapter 4. In Chapter 5, a

discussion of the limitations of this study and direction for future research ends this

thesis.









correlation is excessive, the coefficients and standard errors of the independent

variables become large, making it difficult or impossible to assess the relative

importance of the predictor variables. Using the panel procedure, you do not need to

create dummy variables and compute deviations from the group means. After sorting

the cross-section and year, it will drop the last sub sector and year dummy

automatically. This procedure would report correct MSE, SEE, R2, and standard errors,

and conducts the F test for the fixed group effect as well.









ACKNOWLEDGMENTS

I extend my deepest gratitude to the chair of my supervisory committee, Dr.

Richard Weldon. Special thanks should be given to my committee member Dr. Carlos J.

Trejo-Pech, and Dr. Randle. I acknowledge and express my sincere appreciation for

their direction, assistance, guidance, and for their encouragement to successfully

complete the Master of Science program.









This study provides information regarding the financial performance of

agribusiness. And it advances the understanding of operating efficiency and

effectiveness of agribusiness.










Table 3-4. Summary statistics of explanatory variables
Explanatory Variables GM/S IN/S
Industries SIC Mean Std Mean
Food processing and beverage: 36.49% 3.56% 2.04%
Meat 201 11.93% 1.75% 0.98%
Diary 202 26.28% 1.51% 1.47%
C.F, and PF and V 203 39.18% 3.12% 2.28%
Grain mill 204 39.70% 10.36% 2.05%
Bakery 205 48.58% 3.56% 2.64%
Sugar and confectionery 206 44.26% 4.51% 2.05%
Fats and oils 207 11.91% 2.57% 1.55%
Beverages 208 51.04% 1.95% 2.66%
Miscellaneous food kindred 209 40.26% 2.73% 2.12%
Tobacco 21 51.73% 3.53% 2.59%
Food wholesale, retail, and services: 22.08% 2.21% 1.45%
Food wholesalers 514 14.92% 2.86% 0.60%
Food store-retail 54 25.43% 1.61% 1.14%
Food service 581 25.88% 2.17% 2.62%
Total food supply chain 29.28% 2.88% 1.74%


S,G&A/S
Std Mean Std
0.64% 22.40% 3.00%
0.17% 6.50% 0.85%
0.61% 18.11% 1.34%
0.44% 22.58% 2.24%
0.49% 25.68% 8.69%
2.21% 34.60% 5.61%
0.51% 28.10% 2.15%
0.55% 3.35% 0.80%
0.35% 31.47% 2.22%
0.36% 27.37% 1.92%
0.69% 26.26% 4.21%
0.47% 13.64% 1.47%
0.19% 11.04% 2.00%
0.21% 20.07% 1.12%
1.01% 9.82% 1.30%
0.55% 18.02% 2.24%


R&D/S
Mean Std
0.41% 0.15%
0.03% 0.02%
0.01% 0.01%
0.46% 0.06%
0.81% 0.15%
0.24% 0.35%
0.46% 0.22%
0.11% 0.04%
0.25% 0.09%
0.83% 0.28%
0.87% 0.32%
0.01% 0.01%
0.01% 0.02%
0.00% 0.00%
0.02% 0.01%
0.21% 0.08%


ARTR
Mean Std
11.43 1.49
18.93 2.29
12.22 0.95
9.77 1.15
11.52 0.93
11.76 2.60
8.63 1.13
10.79 2.49
9.30 1.05
8.76 1.02
12.60 1.28
33.89 8.39
19.06 3.10
52.70 15.35
29.92 6.72
22.66 4.94










Table 3-4. Continued
Explanatory Variables
Industries
Food processing and beverage:
Meat
Diary
C.F, and PF and V
Grain mill
Bakery
Sugar and confectionery
Fats and oils
Beverages
Miscellaneous food kindred
Tobacco
Food wholesale, retail, and services:
Food wholesalers
Food store-retail
Food service
Total food supply chain


INTR
Mean
6.62
11.25
12.46
4.30
6.30
7.44
4.59
7.39
5.57
4.03
2.86
18.76
14.47
10.02
31.78
12.69


PP&ETR STD/A CD/A LTD/A
Std Mean Std Mean Std Mean Std Mean Std
1.39 4.21 0.75 1.55% 2.40% 6.27% 3.16% 24.18% 5.97%
2.66 6.67 1.43 1.63% 2.48% 4.39% 2.41% 26.08% 3.54%
2.50 5.29 0.58 0.39% 0.68% 3.42% 2.24% 32.03% 13.60%
0.22 3.84 0.48 3.68% 3.80% 9.96% 4.18% 27.04% 8.11%
0.56 3.79 0.88 2.61% 3.52% 9.70% 2.77% 25.31% 4.18%
4.44 3.31 0.45 0.53% 0.82% 3.31% 3.77% 22.47% 11.09%
0.38 3.54 0.34 0.59% 0.67% 9.20% 3.66% 16.52% 4.22%
1.26 4.50 1.91 0.55% 0.83% 4.20% 4.22% 19.81% 2.98%
0.83 2.74 0.20 1.44% 2.01% 6.85% 2.14% 24.33% 2.05%
0.51 4.11 0.72 1.41% 3.32% 6.99% 4.18% 22.37% 3.20%
0.55 4.34 0.50 2.65% 5.86% 4.74% 2.05% 25.88% 6.71%
1.83 7.83 1.12 0.67% 0.94% 4.29% 1.81% 30.45% 5.43%
1.17 15.57 1.90 1.05% 1.60% 3.44% 1.55% 25.25% 5.88%
0.74 6.25 1.25 0.52% 0.76% 5.83% 2.60% 33.48% 5.39%
3.58 1.67 0.22 0.44% 0.47% 3.59% 1.29% 32.61% 5.02%
1.61 6.02 0.94 1.11% 1.67% 5.28% 2.49% 27.32% 5.70%









The aforementioned ten determinants are disaggregated from the DuPont

framework to measure agribusiness's profitability and perform further financial analysis

(Table 2-1).

EBTOCE in Profitability Measurement

Similar to previous studies, average shareholders' equity is not a stable

measurement due to the net earning fluctuation that can occur over time. In this study,

the earnings before tax on capital employed (EBTOCE) is substituted for ROE. The

calculation of EBTOCE is earnings before tax divided by capital employed. The

EBTOCE is a relative more comprehensive measure of a firm's ability to generate

returns to pay for its cost of capital employed.

Capital employed may be defined in a number of ways. Here, I use the average

net capital employed, which is the summation of fixed assets, investments, and net

working capital. It represents the capital investment necessary for a business to

operate.

This substitution is made for three reasons. First, interest expense is one of the

research concerns in this study, which depicts the management financing strategy and

affects agribusiness's performance. Whereas, as a fixed portion of net income, tax

expense is excluded from the measurement of profitability. Second, the earnings in the

numerator match the range of denominator: capital employed. Because the "capital

employed" includes borrowings, so that the earnings should consider the interest cost to

match the funds sources. Third, as long as EBTOCE is consistently computed for all

firms, it has comparison power to interpret the profitability in terms of ratio EBTOCE.

The overall approach of this study is based on the framework provided by the

DuPont expansion. However, the financial ratios are not exactly derived from DuPont









only difference is that random effects assume the intercept is uncorrelated with each

explanatory variable.

In this study, we assume that the agribusiness profitability is non-randomly

affected by both the cross-section and time-series. Because of the assumption that the

cross-section and time-series effects are fixed, the models are essentially regression

models with dummy variables that correspond to the specified effects. This two-way

fixed effects model is also refers to the two-way least square dummy variable model

(two-way LSDV).

The two-way fixed effects regression model applies to the DuPont components are

as follows:

Yit =ao+3' Xit + ui+ vt+ Eit (2-6)

Where

Yit = EBTOCE of the industry i in the year t,

ao = Intercept coefficient of dropped dummy industry i in the dropped year t,

3' = Slope coefficient of regressors,

Xit = Independent financial ratios of industry i in the year t,

ui = cross-section effects that are constant over time,

vt = time effects that are common to all groups, and

Eit = Residual error for industry i in year t.

This panel regression assumes that slopes are constant, only intercepts vary

according to cross-section and time. This model specifies i-1 sub sector dummies, t-1

time dummies to avoid perfect multicollinearity (the dummy variable trap).

Multicollinearity refers to excessive correlation of the predictor variables. When









multicollinearity (the dummy variable trap), the model drops one cross-section and one

time-series dummy variables in a two-way fixed effect model.

Table 4.3 describes the basic output of the two-way fixed-effects model for the

food processing and beverage industry. Panel A shows that there are ten cross-

sections and twenty three time observations. Ten cross sections refer to the 3-digit SIC

code that starts with 20 and the tobacco sub sector with SIC code 21. The 23 time

observations cover the periods from 1986 to 2008. The F test for the fixed effects,

shown in Table 4-3 Panel B, tests the null hypothesis that there are no fixed effects. The

P value for the F-statistic is almost zero. So we reject the null hypothesis which

indicates that we cannot use OLS to estimate the parameters. There are sub sector

effects, time effects, or both. The test is highly significant. To explore the detail of fixed

effects, two separate one-way fixed effect regressions are developed regarding the sub

sector difference and time impact, respectively. Table 4-3 Panel C provides the results

of two separate F-tests, their small P values are indicating the existence of both sub

sector effects and time effects.

Table 4-4 presents the two-way fixed effects regression results with respect to

FPB. There are ten cross-section and 23 time series in that model. Variables dl, d2 to

d10 are assigned to ten sub sector dummies; (p 1, (p 2 to (p 23 are assigned to the 23

year dummies. SAS sorts the dummy variables by an ascending order and drop the last

cross-section dummy and time dummy automatically. The 230 regression equations

(23x10) can be drawn on the combinations of ten sub sectors and 23 years. The two-

way fixed effect panel regression model assumes that independent variables have

constant slopes, only intercepts vary according to cross-section and time. The









PP&ETR (net sales/ average PP&E) is statistically significant at the 5% level. It means

that one unit of higher PP&E turnover (annual rate in times) increases EBTOCE by 0.01

percent. Keeping net sales constant, the lower investment in PP&E leads to lower

capital employed, which is total assets minus current liability. Thus the higher PP&E

turnover rate, the higher is the EBTOCE.

The final component of the DuPont expansion is Leverage. Under this, three

interest bearing debts are tested: STD/A, CD/A, and LTD/A. The long term debt to

assets ratio (LTD/A) is statistically significant, while the other two are not. LTD/A has a

coefficient of -0.142 and P-value of 0.011 which shows an inverse relationship with

EBTOCE. One percent increase in long term debt results in a decrease in EBTOCE by

0.142 percent. That inverse relationship indicates that the debt to equity capital mix in

the current FPB industry should be reduced to improve return on capital employed.

The results for the food wholesale, retail and service (FWRS) industry are

provided in Table 4-2. The regression model with respects to the FWRS has an R-

squared of 0.7955 and an F-statistics of 9.77 with a probability value of 0.00. It means

that the model measures 79.55 % of EBTOCE's variability. In contrast with the FPB,

only interest expense over sales (IN/S) and long term debt to assets ratio (LTD/A) are

statistically significant at 5% level, with a coefficient of -4.46% and -0.56%. Debt load is

significantly affecting the profitability of FWRS inversely. One percent increase in LTD/A

ratio will result in a decrease in EBTOCE by -0.56% and one percent increase in

interest expense per dollar of sales will decrease EBTOCE by -4.46%.

Besides above ratios, GM/S and INTR are significant at 15% level, and STD/A is

significant at 10% level. It is surprising to see the statistical insignificance of other









To be consistent with other food supply chain studies, the selection of

agribusiness in this research is based on the 3-digit SIC code. Two major industries and

their thirteen sub sectors are considered to comprise the U.S. food supply chain. One

major industry is agricultural processing and marketing (referred to food processing and

beverage, or FPB). This industry belongs to the manufacturing division. The second

industry is agricultural wholesale and retail trade (referred to food wholesale, retail, and

service, or FWRS). This group is classified as part of the wholesale trade and retail

trade division.

The ten sub sectors in the FPB include: meat (SIC 201); dairy (SIC 202); canned,

frozen, and preserved fruits, vegetables, and food specialties (SIC 203); grain mill (SIC

204); bakery (SIC 205); sugar and confectionery (SIC 206); fats and oils (SIC 207);

beverages (SIC 208); miscellaneous food preparations and kindred (SIC 209); and

tobacco (SIC 21). The three sub sectors of the FWRS are; wholesalers (SIC 514),

retailers (SIC 54), and the food service industry (SIC 581). This classification follows

Trejo-Pech, Weldon, and House (2008).

The sample contains 6157 firm-year observations for the 1986-2008 time period.

88 firms had missing values for major financial items and were removed from the

sample. The final sample includes 6069 firm-year observations. 49.11% of the total

observations belonged to the FPB, and 52.79% to the food wholesale, retail, and

services.

Firm Number Fluctuations

Figures 3-1 and 3-2 show the change in the number of agribusinesses in the food

supply chain over the past 23 years. The number of firms fluctuates within the years

according to the business peaks and troughs (business cycles). Table 3-1 summarizes









Table 2-1. Sub components of profitability measurement
Total profitability measurement Components ratios
Profitability EBTOCE


Operating efficiency






Asset use efficiency




Financial leverage


GM/S
SG&A/S
IN/S
R&D/S


ARTR
INTR
PP&ETR


STD/A
CD/A
LTD/A









The second objective of the study is to examine whether the agribusiness's

profitability varies across sub sectors within the U.S. food supply chain. Industrial

economists define an industry as a group of firms that produce homogenous products

such that consumers consider them substitutes. With similar operating features,

agribusinesses in the same sub sector (i.e., same group under 3-digit SIC) might have

similar customer groups, production requirements, and organizational structure. From

the financial viewpoint, those agribusinesses would tend to have comparable profit

margins, asset turnovers and capital structures. The examination of the relationships

between the profitability and cross-section effect leads to my second hypothesis:

Hypothesis 2

The estimators for the coefficients of cross-sectional effects are not jointly

significant. Profitability does not have cross-sectional effects within the U.S. food supply

chain. All sub sectors' profitability levels have no significant difference statistically.

The final objective of this research is to test whether there is a time fixed effect in

the agribusiness's profitability. Are there time effects that are common to all sub

sectors? For example, do all sub sectors experience a decrease in return in a specific

recession? Some economists are of the view that only sales of luxury and non-essential

goods will be significantly affected by recessions, while necessary goods like food,

water, and clothing will not be affected as severely. The food industry generally supplies

necessities and food consumers are very reluctant to forgo consumption even during

the hard times. While they might substitute margarine for butter, people will still need to

satisfy their basic demands. This study explores whether the profitability fluctuates

along with the economic downturns. Additionally, it evaluates the agribusinesses'









BIOGRAPHICAL SKETCH

Youshan Zhao was born in Beijing and obtained her bachelor's degree in

accounting from North China University of Technology, Beijing, China in 2003. Miss.

Zhao is China Certified Public Accountant and worked for

PRICEWATERHOUSECOOPERS Beijing, China for three and a half years. The

requirements for the degree of Master of Science in food and resource economics with

a minor in statistics were completed in August 2010 at the University of Florida.









Table 3-4 provides the mean and standard deviation of the ten explanatory

variables by all sub sectors among food supply chain over the 23 years. Overall, in

terms of gross margin rate, FPB industry outperforms the FWRS with 36.49% (standard

deviation of 3.56%) compared to 22.08% (2.21%). The most profitable sub sectors in

the food supply chain, like tobacco, sugar and confectionery, and beverage are all

components of FPB. Besides, FPB operates more "aggressively" than FWRS by

spending a larger portion of net sales on selling expense, G&A expense, interest

expense, and R&D expenditure. From the perspective of assets use efficiency,

companies with low profit margins usually tend to have high asset turnover. This is true

in food supply chain as well. FWRS has the higher assets turnover rate (means faster in

times per year) than FPB. In comparison, the accounts receivable turnover rate in

FWRS industry (33.89) is 2.97 times higher than FPB (11.43). That might be because

the trading and services industry are dealing with final customers so that it involves less

receivables. The inventory turnover rate in FWRS (18.76) is 2.83 times faster than FPB

(6.62), indicating that FWRS has better inventory management than manufactures in

FPB. PP&E turnover rate in FWRS (7.83) is 1.86 times more than FPB (4.21), because

the former industry requires less manufacturing equipment than the latter. The above

indicates that the FWRS has higher assets usage efficiency in generating sales. With

respect to capital structure, FPB has a similar portion of short-term debt mixture but a

slightly larger scale of long-term debt mixture. It is worth to mention that long-term debt

is the main financing source for agribusiness. For the entire food supply chain, the

short-term debt (exclude current-portion of long-term debt) is only 1.11%, compared

with a long-term debt to assets ratio of 27.32%.












0.50

0.45

0.40

0.35

0.30

0.25

0.20

0.15

0.10

0.05

0.00
1984


1996 1998 2000 2002 2004 2006


- SIC 208 Beverage SIC 209 Miscellaneous Food SIC 210Tobacco


Figure 3-4. Continued


1986 1988 1990 1992 1994


SIC 206 Sugar and confectionery SIC 207 Fats and oils









readily than a firm that bears no fixed debt cost. In industry downturns, the highly

leveraged firms lose substantial market share to their more conservatively financed

competitors. There is a positive relationship between financial condition and firm

performance (Opler and Sheridan, 1994). Considering the importance of leverage, this

study will examine the agribusiness' total profitability decomposing all three DuPont

components (PM, ATO and Leverage) and their underlying ratios.

The main drawback of the DuPont system, as claimed by Banker, Chang, and

Majumdar (1993), is that this system and its underlying ratios provide only a gross

aggregate measure for firms' profitability. This system doesn't easily capture the micro-

attributes of productivity of firms, like product mix, price recovery, capacity utilization,

etc.

With respect to agricultural sectors, the economics literature provides multiple

explanations for the profitability measures. Forster (1996) uses the capital asset pricing

model (CAPM) to investigate the rate of return, solvency, liquidity and other financial

ratio measurements in agribusinesses. Foster found that the capital structure and

business risk of agribusiness are interrelated. Both of them are important determinants

of stock returns. In addition to the specific within firm financial factors mentioned above,

there is an increasing interest regarding agribusiness profitability in response to

macroeconomic conditions. Empirical results indicate that macroeconomic conditions

(i.e. fiscal policy, business cycles) have differing effects on agribusiness profitability

dependent on a firm's financial structure and their market segment (Neibergs, 1998).

Regarding the agribusiness sector, the financial ratio analysis in this study follows

from recognition of accounting ratios relationships that determine total profitability.









strategy. Economic recessions do not have significant impact on profitability of the U.S.

food supply chain statistically.

The major research limitation of this study was the failure to collect the market

variables, such as stock prices, price earnings rates (PE) and earnings per share (EPS).

These are indicators of a company's profitability as perceived by investors. That is also

the drawback of DuPont formula, which is the main focus on the internal measurement

of firm's performance.









spending is generally a sign of a recession, which is caused by a mismatch of

aggregated demand and aggregate supply.

The fluctuation of production or economic activity between periods is described as

a business cycle. It has a sequence of four phases: peak, contraction, trough, and

expansion. A peak is defined as the time period when business activity is at its

maximum. However, this maximum point does not last forever, and business activity

starts to decline and contract. A trough is the time period when business activity

reaches a minimum. However, business activities start to pick up, this is the

expansionary period.

Since 1854, the U.S. has encountered 32 cycles of expansion and contraction.

Significant amount of research has been done to explain the relationship of recession

and aggregate demand changes. In this study, however, the impact of the recent three

main recessions on the supply side is examined.

The three major recessions considered in this research are: 1) July 1990-March

1991: 8 months (the early 1990s recession), 2) March 2001-November 2001: 8 months

(the early 2000s recession), and 3) December 2007-to date: more than 25 months (the

late 2000s recession).

From the demand side, a recession reduces consumer's ability and confidence to

spend. The consumers, who overspent based on credit secured by bubble priced

assets, drove the U.S. economy. With the decline in the bubble assets, consumers'

willingness to spend was hit hard, shaking their spending confidence. Industry was

adversely affected as sales that depend on consumer's consumption fell. The effect of a

recession was catastrophic: GDP, employment, investment spending, capacity









TABLE OF CONTENTS

page

A C KN O W LED G M ENTS ............................ ............... ......................................... 4

L IS T O F T A B L E S ............................................................................................................ 7

LIS T O F F IG U R E S .................................................................. 8

A B S T R A C T ..................................................................................................................... 9

CHAPTER

1 IN T R O D U C T IO N .................................................... .......... 11

Agribusiness Profitability ................ ..... ... ......... ..... 11
Agribusiness and the U.S. Food Supply Chain ....... ........ ...... .................. 12
Economic Recession ...... ...................... .................. 12
Research Objectives ...... .................... ......... .......... 14
Test Hypothesis 1... ........ ................... ............... 14
Test Hypothesis 2................................ ............... 15
Test Hypothesis 3................................ ............... 16

2 LITERATURE REVIEW AND RESEARCH METHODOLOGY.......................... 17

Literature Review ............... ........ .................. 17
Total Orofitability Measurement ................. ... .............................................. 20
Basic DuPont Formulation ...... ............ ............................. 20
Expansion of DuPont Formula............................. 20
Decomposition of Profit Margin ................. ................. .......... 21
Decomposition of Assets Turnover ....... ........ ........... ................ 21
Decomposition and Substitution of Leverage....................... 22
EBTOCE in Profitability Measurement ............. ......... ......... ............ 25
Statistical Model...................... ....... ............... 26
Basic Regression Model............................... .................... 26
Panel Data................................ ............ ............... 27

3 DATA SOURCES AND DESCRIPTION....................................... 32

D a ta a n d T re n d s ............................ .. ......... .... .. ................................. ............... 3 2
Data Sources..... ....................................... ............... 32
Firm Num ber Fluctuations............. .. ....... ....................... ... ..... .......... 33
W weighted Average M ethod ........ ............................................... ................. 34
Summary Statistics and Accounting Variables............................ ............... 35









where STD/A or Short-term borrowings (including short-term loan and short-term note-

payable) to assets = (short-term borrowings)/ (total assets); CD/A or current portion of

long-term debt to assets = (current portion of debt)/ (total assets); Long-term debt to

assets (LTD/A) = (long-term debt)/ (total assets); and E is residuals. Note that the

current portion of long-term debt (CD) is presented separately with the short-term debt

in this study for the reason that these two types of debt have different interest rates and

are representing different financing strategy of firm. The CD is also separated from the

long-term debt because it reflects the payment that is due within a year, which might be

an element that pushes the management to take action to improve liquidity.

Many studies isolate the financial structure from the DuPont decomposition as a

separate research field. They believe that unlike PM and ATO, leverage is a decision

that is highly controlled by management. Some profitability valuation studies ignore the

leverage factor. Authors believe the firm's profitability is mainly driven by PM and ATO

(Fairfield and Yohn 2001; Nissim and Penman 2001; Penman and Zhang 2003;

Fairfield, Sweeney and Yohn). These studies argue that the financial structure and

associated return should be ignored when performing profitability analysis, because it

can be manipulated by management discretionally. In my study, I include the financial

leverage ratio to measure the total profitability. The financial structure captures the

firm's ability to use available economic resources to increase profit. This phenomenon is

often referred to as the multiplier effect because the way of financing of the firm can

affect the value of shareholder's equity.









Table 3-2. EBTOCE of FWRS industry by year and sub sector
Sector Retailers Food wholesalers Food service
SIC 54 514 581 Mean


1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
MEAN
Std


14.09%
16.82%
11.34%
10.27%
12.45%
12.38%
10.02%
10.27%
9.09%
12.03%
11.86%
11.48%
10.52%
11.49%
8.18%
8.28%
6.32%
3.97%
2.48%
6.66%
9.57%
10.89%
5.23%
9.81%
3.28%


12.45%
11.78%
9.94%
10.69%
-0.50%
10.95%
10.55%
10.26%
10.90%
10.71%
9.45%
9.79%
0.76%
9.70%
7.48%
16.54%
24.43%
24.24%
25.08%
25.52%
20.41%
22.53%
24.96%
13.85%
7.64%


7.23%
11.01%
10.89%
9.11%
8.23%
9.26%
8.86%
-3.86%
11.17%
8.35%
8.36%
7.36%
14.26%
11.94%
13.76%
13.08%
11.78%
11.14%
12.59%
14.12%
15.11%
15.27%
12.56%
10.50%
3.98%


11.26%
13.20%
10.72%
10.02%
6.73%
10.86%
9.81%
5.56%
10.39%
10.36%
9.89%
9.54%
8.51%
11.04%
9.81%
12.63%
14.18%
13.12%
13.38%
15.43%
15.03%
16.23%
14.25%
11.39%
2.70%










Firm number in the FPB industry


1986 1987 1988 1989 19901991991 1992 1993 1994 1995 1996 19971998 1999 2000 2001 2002 2003 2004 2005 2006 20072008
Firm Number 116 124 122 115 115 116 119 138 147 156 166 163 155 145 131 115 111 110 105 102 97 102 95
Figure 3-1. Firm number in FPB industry










Panel Data

Panel data sets usually include multiple cross-sectional data (i.e. country, blocks,

and firms) which are observed over two or more time periods. The characteristic of

panel data was summarized by Yaffee, (2003), "Panel data analysis endows regression

analysis with both a spatial and temporal dimension." Panel data has become widely

analyzed to understand social and economic changes over a time span. When both the

space and time span is considered, the panel data analysis provides a number of ways

to improve the interpretation of data.

The OLS regression assumes that the residuals are independent. However, in a

typical panel data set, the residual components E, in (2-5), are highly likely to be

correlated both with the cross-section error and the time-series errors. That violates the

assumptions of OLS regression and can lead to biased estimates of coefficients and

biased estimates of the standard errors. To avoid that common problem of the OLS

model, a two-way fixed effects panel model is used in this research, which assumes that

the error structure is corresponding to both cross-section and time effect

simultaneously.

The fixed effects model is one of several types of panel analytic models. If the

specification is dependent only on the cross section to which the observation belongs,

such a model is referred to as a one-way model. When a data structure depends on

both the cross sections and the time periods to which the observation belongs, it is

called a two-way model. Additionally, if the data structure provides non-random effect

pattern, it is called fixed-effects model. Otherwise it is the random-effects models. The










Table 4-3. Continued
Panel B : Fixed effect F test


Model Fixed effect Sample Size F Test Prob value Result
Two-way Both sectors and time: 230 F(31,186)=5.15 <.0001 Null hypothesis for
I =201, 202, 203..., 21 both sector and time
T =1986, 1987, effect are not
1988.....,2008 rejected
Table 4-3 provides the regression results of cross-section effects and time effects in model: EBTOCE(it) = a0+31
GM/S(it)+ 32 SG&A/S(it)+ 33 IN/S(it)+ 34R&D/S(it)+ 35 ARTR(it)+ 36INTR(it) + 37PP&ETR(it)+ 38STD/A(it)+ 39CD/A(it)
+ 310LTD/A(it)+ ui+ vt +E(it).

































2010 Youshan Zhao










Table 2-2. Variable definition
Abbreviation Description
ROE Return on equity

ROA Return on assets

EBT Earnings before tax

EBTOCE Return on capital employed

PM Net profit margin rate

GM/S Gross margin rate

SG&A/S Selling,general and
administrative expense rate
IN/S Interest expenses rate


R&D/S


ATO

NOATO

ARTR

INTR

PP&ETR

STD/A

CD/A

LTD/A


Research and development
expense rate

Total assets turnover

Net operating assets turnover

Accounts receivables turnover
rate
Inventory turnover rate

Property, plant and equipment

Short term debt to assets ratio

Current portion of long-term debt
to assets ratio
Long-term debt to assets ratio


Formula
Net income/ equity

Net income/ total assets

Pretax income

Earnings before tax / capital employed

Pretax income/ net sales

(Net sales-COGS)/ net sales

General and administrative expense / net
sales
Interest expenses/ net sales;

Research and development expense/ net
sales.

Net sales/average total assets

Net sales/average net operating assets

Net sales/ average accounts receivable

COGS/ average inventory;

Nets sales / average property, plant and
equipment.
Short term debt/ total Assets

Current portion of long-term debt/ total
assets
Long-term debt/ total assets









CHAPTER 2
LITERATURE REVIEW AND RESEARCH METHODOLOGY

This chapter will first provide the reader with a review of literature. The next

section explores the DuPont decomposition and measurement proxies for the

regression analysis. Finally, the foundation of the two-way fixed effect regression is

introduced. Both tools will be used to derive the empirical results based on the

agribusiness financial data.

Literature Review

Financial ratios analysis within an industry peer group has been traditionally

viewed as a major method that compares and measures firms' performance in the

competitive market. A number of historical accounting ratios are viewed as information

that can forecast performance. By measuring the typical values and patterns for ratios

over time, financial statement analysis is presented as a matter of pro forma analysis of

the future (Nissim and Penman, 2001).

Among all the financial performance evaluation models, the DuPont system is the

most common used for providing fundamental analysis regarding the total profitability of

the firm. The interest in the DuPont system and profitability drivers' analysis is

developing the association between current and future financial ratios. It is suggested

that by widening the accounting information set, profitability measures, such as return

on equity (ROE), will exhibit greater predictive power. The firms' time series behavior

and difference in ratios across firms at a given time can be reflected in the changes in

earnings (Freeman and Penman, 1982). As two multiplicative components of DuPont

analysis, profit margin (PM) and total asset turnover (ATO) have been extensively

examined by decomposing from the return on total assets (ROA), which is a sub









Firm number in the FWRS industry


1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008
-- Firmnumber 157 156 142 128 126 132 148 167 175 187 203 193 168 154 143 122 118 112 104 102 100 89 78
Figure 3-2. Firm number in the FWRS industry


ML


'Ooooow









LIST OF REFERENCES


Banker, R., Chang, H. S., and Majumdar, S. K. 1993. "Analyzing the underlying
dimensions of firm profitability. "Managerial and Decision Economics, 14: 25-36.

Bourlakis, M. A., and Weightman, P. W. 2004. Food Supply Chain Management. Wiley-
Blackwell.

Fairfield, P., J. Whisenant, and T. Yohn. 2003. "The differential persistence of accruals
and cash flows for future operating income versus future profitability." The Review of
Accounting Studies 8: 221-243.

Forster, D. L. 1996. "Capital structure, business risk, and investor returns for
agribusinesses." Agribusiness, 12: 429-442.

Freeman, R. J., and Penman, S. 1982. "Book rate-of-return and prediction of earnings
changes: An empirical investigation." Journal of Accounting Research, 20: 639-653.

Jensen, M. C. 1989. "The eclipse of the public corporation". Harvard Business Review,
5,61-74.

Magdoff, H., Foster, J. B., and Buttel, F. H. 2000. Hungry for Profit: The Agribusiness
Threat to Farmers, Food, and the Environment. Monthly Review Press.

Neibergs, J. S. 1998. "Macroeconomic Conditions and Agribusiness Profitability: An
Analysis Using Pooled Data." International Food and Agribusiness Management
Review, 1: 91-105.

Nissim, D. and S.H. Penman. 2001. "Ratio analysis and equity valuation: From research
to practice." Review of Accounting Studies, 6: 109-154.

Opler, T. C., and Sheridan, T. 1994. "Financial Distress and Corporate Performance."
The Journal of Finance, 49: 1015-1040.

Selling, T., and Stickney, C. 1989. "The effects of business environments and strategy
on a firm's rate of return on assets." Financial Analysts Journal, 45: 43-52.

Trejo-Pech, C. O., Weldon, R. N., and House, L. A. 2008. "Earnings, accruals, cash
flows, and EBITDA for agribusiness firms." Agricultural Finance Review, 68: 301-
319.

Yaffee, R. 2003. "A primer for panel data analyses."
http://www.nyu.edu/its/pubs/connect/fall03/yaffee primer.html, last accessed July
2010.









the decrease of market players during economic downturns. For example, the average

number of firm over the 23 years for the FPB is 125. The number of firms decreased by

ten during the early 1990s recession (calculated as the average number of firms in the

time period minus number of firms in 1990). Table 3-1 shows that the number of firms in

each recession period is apparently lower than the average firm number over the past

23 years. Since it is difficult to get the firm number for the specific month during the

recession, the number of firm number at the end of 1999 is used to represent the firm

number in the early 1990s recession. Similarly, firm numbers at the end of 2001 and

2008 are used to represent the early 2000s recession and the late 2000s recession

respectively.

Weighted Average Method

In every stage of an industry, new firms enter while distressed firms leave. This

causes the panel data to be variable and unbalanced. Additionally, particularly for the

agribusiness sample, some firms operated for only a few periods during the past 23

years and then were merged or acquired. This is another reason for the unbalanced

nature of the data set. Considering that the purpose of this study is to explore the sub

sectors effects and time effect on the sub sectors' EBTOCE as opposed to the

individual firms', the sub sector average is obtained to perform the two-way fixed effect

panel regression.

The majority of past research employs the arithmetic mean of financial ratios as

proxy of industry performance. But the straight average value does not reflect the

industry objectively since it assumes that all firms have equal influence on that industry.

It is felt that a weighted mean of financial ratios to measure the industry's primary

financial condition is a better proxy of industry performance. In a competitive market,









Table 4-4. Continued
Parameter Estimates


Variable Variable Estimate Standard Error t Value
Year 1991 (p 6 0.053 0.020 2.690
Year 1992 (p 7 0.044 0.019 2.270
Year 1993 (p 8 0.035 0.019 1.810
Year 1994 (p 9 0.037 0.019 2.010
Year 1995 (p 10 0.056 0.019 2.970
Year 1996 p 11 0.040 0.019 2.150
Year 1997 (p 12 0.024 0.018 1.340
Year 1998 (p 13 0.014 0.018 0.760
Year 1999 (p 14 0.045 0.019 2.390
Year 2000 (p 15 0.025 0.019 1.320
Year 2001 (p 16 0.020 0.018 1.080
Year 2002 (p 17 0.017 0.018 0.960
Year 2003 (p 18 0.001 0.018 0.030
Year 2004 (p 19 0.008 0.018 0.440
Year 2005 (p 20 0.018 0.018 1.030
Year 2006 (p 21 0.003 0.018 0.180
Year 2007 (p 22 0.021 0.017 1.200
Sector 21 in 2008 d10(p 23 -0.038 0.056 -0.690
Ten explanatory variables parameter estimates are shown in table4-1


Pr > Itl
0.008
0.024
0.072
0.046
0.003
0.033
0.183
0.448
0.018
0.190
0.282
0.339
0.976
0.659
0.302
0.860
0.233
0.494


Label
Time Series
Time Series
Time Series
Time Series
Time Series
Time Series
Time Series
Time Series
Time Series
Time Series
Time Series
Time Series
Time Series
Time Series
Time Series
Time Series
Time Series
Intercept


SIC groups classifications follows: Meat (201); diary (202); canned, frozen, and preserved fruits and vegetables (203); grain
mill (204); bakery (SIC 205); sugar and confectionery (206); fats and oils (207); beverages (208);miscellaneous food
preparations and kindred (209); tobacco (21); food service (5810 and 5812); retailers (5400 and 5411); and wholesalers (5140,
5141, and 5180).


Effect
Effect
Effect
Effect
Effect
Effect
Effect
Effect
Effect
Effect
Effect
Effect
Effect
Effect
Effect
Effect
Effect









LIST OF TABLES


Table page

2-1 Sub components of profitability measurement........................ ................. 30

2-2 Variable definition ................ .............. ............ ............. ............... 31

3-1 Firm number change during the past three recessions................. .......... 41

3-2 EBTOCE of FWRS industry by year and sub sector................ .................... 42

3-3 EBTOCE of FPB industry by year and sub sector .................... .................. 43

3-4 Summary statistics of explanatory variables.................. ......................... 44

4-1 Two-way fixed effects panel regression in FPB industry ......................... 59

4-2 Two-way fixed effects panel regression in FWRS industry.............................. 60

4-3 Hypothesis test for fixed cross-section effects and time effects in the FPB........ 61

4-4 Parameter estimates in the fixed effects model for the FPB............................. 64

4-5 Hypothesis test for the fixed cross-section effects and time effects in the
FW RS industry ............... ... ......... ............... .. ..................... .......... 66

4-6 Parameter estimates in the fixed effects model for FWRS industry.................. 69









Besides, using the cross-section time series data, we investigate how earnings

components exhibit the time-series properties and cross-section properties.

Total Profitability Measurement

Basic DuPont Formulation

ROE measures the rate of return on the ownership interest in a firm. It is

measured as net income (NI) divided by average total equity (ATE). The DuPont

Company started using the breakdown of ROE to measure efficiency at generating

profits from every unit of shareholders' equity. Basically, the ROE can be decomposed

into three parts. The following decomposition, also called the DuPont expansion, is

known as a strategic profit model. Mathematically:



Net income Sales Total Assets
ROE = x x
Sales Total Assets Average stockholder equity (2-1)



The first ratio in 2-1 refers to the PM; the second ratio refers to ATO, and the third

ratio of the DuPont formula reflects the leverage of the firm. Multiplying the first two

fractions (PM*ATO) will yield the return on assets (ROA), which provides a picture of the

ability of using assets to earn a profit.

Expansion of DuPont Formula

Three ratios are expanded from the DuPont system to measure ROE; PM, ATO

and leverage. Those financial ratios will be decomposed to the next level as micro

proxies of firm operation conditions. The major variables substitutions are explained in

the following sections.









is not statistically affecting the EBTOCE for almost all sub sectors in the food supply

chain. In other words, the profitability change, in terms of EBTOCE, cannot be explained

by the variations of assets turnover rate.

In terms of the second and third hypotheses, the two different industries have

different test results. Significant cross-section effects are present in the FPB industry. It

indicates that the profitability is significantly different across sub sectors in that industry.

However, FWRS's profitability is less relevant to its sub sectors, because the cross-

section effect is not reported from the panel data regression statistical test.

It is shown that the economic downturns do have an impact on the number of

market participants. The reduced number of market participants is highly associated

with economic recessions. Especially small agribusinesses, which drop out the market,

file for bankruptcy or merge. However, in this model, the time effect has been tested on

surviving firms. Even though there are significant time effects exhibits in the FPB, the

fluctuation of EBTOCE is not associated with recession periods, which indicates that

there are other factors or reasons that are playing a more important role in the

fluctuation of firm's profitability. EBTOCE haven't been affected materially by the well

recognized recession periods. A decrease in profitability might stem from various factors

or reasons. By analyzing the individual firms' performance closely, it is found that

decreases in return are mainly attributed to ineffective or inefficient management, such

as fail to control manufacturing cost, bad acquisition, failure to hedge the foreign

exchange risk, stronger competitors, etc. Ultimately, this study concludes that the

profitability of the food supply chain is mainly affected by management operating









large firms tend to achieve economies of scale to gain competitive advantage.

Emphasizing the resource efficiencies, productivity, and product quality, large firms

dominate the development of industries. This can be justified because even though an

industry may have many market players of all sizes, the market share is usually

dominated by a single large firm or a few large firms. In my model, the net sales variable

is applied as the base to calculate the weighted average.

The final sample includes 6069 firm-year observations. After taking the weighted

method to calculate the sector average, there are 299 (or 13*23) sector-year samples in

the data set. Each variable in each sub sector is calculated as an average value yearly.

By industries, there are 230 and 69 data lines in FRB and FWRS respectively after

taking the weighted average by net sales. The unbalanced firm-year panel data is

transformed into a balanced sub sector-year panel data. Tables 3-3 and 3-4 report the

summary statistics based on weighted average method regarding the sub sectors over

23 years.

Summary Statistics and Accounting Variables

Summary statistics for all dependent and independent variables are provided in

the Tables 3-2 to 3-4 based on 3-digit SIC sub sectors and their respective industry

groups. Mean values are weighted by net sales. EBTOCE of the two industries are

presented by sub sectors and year in the Table 3-2 and Table 3-3, respectively. For a

better understanding of the EBTOCE fluctuation during the period of 1986-2008,

Figures 3-3 and 3-4 present the corresponding graphs for EBTOCE. Mean and standard

deviation of the ten explanatory ratios for all sub sectors in two industries of food supply

chain are shown by pooled years in Table 3-4.









Table 4-5. Continued
Panel C: separate F-test for individual effect
Model Fixed effect Sample Size F Test Prob value Result
Single sectors effect Null hypothesis for sector
One-way I =54,514,581 69 F(2,55)= 1.57 0.2167 effect is not rejected
Single time effect Null hypothesis for time
One-way T =1986, 1987, 1988.....,2008 69 F(22, 35)=1.21 0.3079 effect is not rejected
Table4-5 provides the regression results of cross-section effects and time effects in model: EBTOCE(it) = a0+31
GM/S(it)+ 32 SG&A/S(it)+ 33 IN/S(it)+ 34R&D/S(it)+ 35 ARTR(it)+ 36INTR(it) + 37PP&ETR(it)+ 38STD/A(it)+ 39CD/A(it)
+ 310LTD/A(it)+ ui+ vt +E(it).









Table 3-1. Firm number change during the past three recessions
Entire food supply chain
FPB FWRS
Firm number Mean Reduced Firm number Mean Reduced
1990 115 125 10 126 139 13
2001 115 125 10 122 139 17
2008 95 125 30 78 139 61









CHAPTER 5
CONCLUSION

This research aims to examine the U.S. agribusiness's profitability based on the

decomposition of the DuPont system with consideration of the recent three U.S.

economic recessions. The overall approach is to derive the financial variables that stem

from the DuPont formula as independent variables, to measure the dependent variable

which is the return on capital employed for agribusinesses. The effects of different

financial aspects to EBTOCE are tested by this regression. Additionally, the cross-

sectional effects and time effects of the U.S. food supply chain over past 23 years are

tested. Through the analysis of profitability fluctuation, this study provides an insight into

the agribusiness financial performance.

This research is a financial statement analysis based on the DuPont system with

respect to U.S. food supply system. Three hypotheses are tested with respect to two

major food industries and subordinate 13 sub sectors.

In terms of the first hypothesis test, the empirical result indicates that agribusiness'

profitability among all sub sectors is negatively affected by long-term debt. The higher

the long term debt to asset mix, the lower return on capital employed. That conclusion

can be applied to both FPB and FWRS. Furthermore, the profit margin (including four

major components: GM/S, SG&A/S, IN/S, R&D/S) have significant impact on the

profitability of all the sub sectors in the FPB industry. However, except for the IN/S,

other profit margin components which are proxies of the firm's operating efficiency

exhibit statistical insignificance in the FWRS industry. Assets turnover ratios have three

major components: ARTR, INTR and PP&ETR. Except the PP&ETR in FPB, coefficient

estimations of other assets turnover ratios provide evidence that assets usage efficiency

































To Jingqi Wang, Zhenya Zhao and Qipeng Zheng









LIST OF FIGURES

Figure page

3-1 Firm Number in FPB Industry .................. ............ ... ............... 46

3-2 Firm Number in the FWRS Industry........ ........................... ............. 47

3-3 EBTOCE Fluctuation by Weighted Average in the FWRS Industry.................... 48

3-4 EBTOCE Fluctuation by Weighted Average in the FPB Industry.................... 49









Table 4-5. Hypothesis test for the fixed cross-section effects and time effects in the FWRS industry
Panel A: Model Description
Estimation Method Fix Two
Number of Cross Sections 3
Time Series Length 23









The average profitability measured in terms of EBTOCE for the entire food

wholesale, retail, and services over past 23 years was 11.39% with a standard deviation

of 2.70%. Breaking down the three sub sectors, the food wholesale sub sector has the

highest profitability of 13.85 %, followed by food services with profitability of 10.5 % and

finally the food retail sub sector with profitability of 9.81%. Their standard deviations

have the same order with EBTOCE. The food wholesale sub sector is the most variable

and food retail sub sector is relatively stable in return rate.

For a better comparison, ten sub sectors within the FPB industry are classified into

two groups according to their SIC order. Each group presents 5 sub sectors. Figure3-4

shows yearly EBTOCE fluctuations for these ten sub sectors respectively. The EBTOCE

in FPB (Table 3-3) by year shows that all ten sub sectors have higher than 10 %

profitability in terms of EBTOCE. The average EBTOCE of the entire FPB during the

past 23 years is 15.48% with a standard deviation of 2.33%. Average EBTOCE for the

FPB is 4.09% higher than the FWRS. Among the ten sub sectors of FPB, sugar and

confectionery is the most profitable sub sector with an average EBTOCE of 21.16%,

followed by the canned, frozen, and preserved fruits, vegetables, and food specialties

(C,F, and PF and V) and grain mill with EBTOCE of 19.86% and 19.31%, respectively.

The diary sub sector is more variable in profitability in terms of EBTOCE during the time

period, shown by a standard deviation of 6.83 %. Furthermore, all sub sectors except

the tobacco and "fats and oils" sub sectors have relatively slightly downward trend

meaning that the average EBTOCE has been decreasing over the years. One possible

explanation of falling profits is that both competition and technology have been

increasing steadily which is shrinking profits.









component of ROE. In particular, the common form to analyze ROA is performed by

focusing on the return on net operating assets (RNOA). RNOA can be decomposed into

PM and net operating assets turnover (NOATO)1. Previous research has found that

there is a negative convex relationship between the level of PM and NOATO. This

property tends to cluster by industry and shows stronger when applied to industry

averages (Selling and Stickney, 1989). In a view of cross-section industries, pooled

firms (regardless of the industries that they belong to) tend to achieve similar levels of

RNOA by varying the combination of PM and NOATO. Furthermore, prior studies

explore the interpretation power of the changes in PM and NOATO metrics at the same

time. Empirical results show that NOATO has more persistent predictive power than

PM. That is, the change in NOATO explains a larger portion of the change in RNOA

(Fairfield and Yohn 2003).

Nissim and Penman (2001) suggest a residual-income valuation framework

without considering the effects of financial leverage. It is commonly believed that the

firm's choice of capital structure can be manipulated. Relevant studies isolate the

leverage and associated returns as discretional attributes by management but not a

desirable operating profitability variable. However, the determination of firms' leverage

rate affects their residual returns. Jensen (1998) argues that the cost of leverage, both

direct cost and indirect cost, might improve corporate performance by forcing

managements to make value-maximizing strategy to avoid debt pressure. For example,

a firm with debt cost from borrowings might cut its underperforming production line more


1 Net operating assets are businesses operating assets minus its operating liabilities. It has been
analyzed widely by separating operating activities from financing activities. The purpose of doing that is to
evaluate firm's operating performance independently.









borrowers) in generating sales revenue. For an agribusiness, the major asset items

include accounts receivables; inventory; property, plant and equipment. Thus, those

asset turnover ratios play the major role in the determination of total asset turnover. A

company may have other forms of assets, like cash and security investment, which are

generally not as important as those mentioned. Those major asset items have more

important commercial substance in explaining the assets usage performance. Similar to

PM, the following expression is used to estimate total asset turnover rate:

ATO=f (ARTR, INTR, PP&ETR, E), (2-2)

where ARTR or accounts receivables turnover rate = (net sales)/ (average accounts

receivables); INTR3 or inventory turnover rate= (COGS)/ (average inventory); PP&ETR

or property, plant and equipment turnover rate = (nets sales) / (average property, plant

and equipment); and E refers to residuals.

Decomposition and Substitution of Leverage

The third part of the DuPont formula is leverage, which is defined as total assets

divided by average equity. Leverage reflects the financial structure of the company and

it is highly influenced by the different business environments. Under economic

downturn, firms might have to borrow more funds to keep operating. Simultaneously,

the financial institutions usually reduce credit which makes it even harder for firm to

borrow money during downturns. The leverage ratio will provide a good explanation on

the influence of capital structure to a firm's profitability. Firms usually use debt to

supplement owner's investment and increase gains. However, these liabilities have

fixed payment obligation on a regular basis (i.e., interest expense). Different firms have

3 Some practitioners prefer using net sales instead of COGS to calculate the INTR. This study adopts
COGS because inventories are usually recorded at cost while net sales are recorded at market value.









CHAPTER 3
DATA SOURCES AND DESCRIPTION

Data and Trends

Data Sources

The financial statement data for this study is derived from the CRSP/Compustat

Merged Database (CCM), accessed from the Wharton Research Data Services

(WRDS). The financial data of the U.S. agribusinesses food supply chain obtained

covers 23 years from 1986 to 2008. During the 23 years, the U.S. went through three

economic recessions: the early 1990s, the early 2000s, and the late 2000s recession.

The major financial items used to calculate the financial ratios, the variables of main

interest in this study, are drawn from the firms' annual income statements and annual

balance sheets.

The Economic Research Service (ERS) is a primary source of economic

information and research in the U.S. Department of Agriculture. The ERS classifies the

farm and farm-related industries based on Standard Industry Classification (SIC) codes.

According to the ERS, there are six major industry groups which satisfy the demand for

agricultural products. These six major groups are:

Farm production,

Agricultural services, forestry, and fishing,

Agricultural inputs industries,

Agricultural processing and marketing,

Agricultural wholesale and retail trade, and

Indirect agribusinesses.









practice or a specific time period that are influencing those small businesses

persistently.









Table 4-1. Two-way fixed effects panel regression in FPB industry
Model Description and Estimation
Total sample 230 Degree of 186
freedom
Fit Statistics (F-statistics: 19.23, p-value: 0.000)
R-Square 0.7182 MSE 0.0013
Parameter Estimates
Estimate Standard Error t Value Pr > Itl
Intercept -0.038 0.056 -0.690 0.494
GM/S 0.798 0.113 7.040 <.0001
IN/S -1.488 0.508 -2.930 0.004
SG&A/S -0.581 0.141 -4.110 <.0001
R&D/S -2.329 0.990 -2.350 0.020
ARTR -0.002 0.002 -0.810 0.419
INTR 0.001 0.002 0.390 0.698
PP&ETR 0.010 0.004 2.760 0.006
STD/A -0.013 0.125 -0.100 0.921
CD/A 0.055 0.099 0.560 0.579
LTD/A -0.142 0.055 -2.570 0.011
Table 4-1 presents the statistical output of the regression for the independent variables in FPB. Since the hypothesis 1, 2, and 3 are
tested separately, the two-way fixed effects F test and results are moved to Table 4-3. The regression takes the form that
EBTOCE(it) = a(i)+31 GM/S(it)+ 32 SG&A/S(it)+ 33 IN/S(it)+ 34R&D/S(it)+ 35 ARTR(it)+ 361NTR(it) + 37PP&ETR(it)+ 38STD/A(it)+
39CD/A(it) + 310LTD/A(it)+ ui+ vt+ E(it). Independent variables are defined as following: GM/S is gross margin rate; SG&A/S is
selling and general and administrative Expense rate; IN/S is Interest Expenses rate; R&D/S is research and development expense
rate; ARTR is accounts receivables turnover rate; INTR is inventory turnover rate; PP&ETR is property, plant and equipment turnover
rate; STD/A, CD/A and LTD/A represent short term debt, current portion of long-term debt and long-term debt to assets ratio
respectively.









Table 4-4. Parameter estimates
Parameter Estimates
Variable Variable
Sub Sector 201 dl
Sub Sector 202 d2
Sub Sector 203 d3
Sub Sector 204 d4
Sub Sector 205 d5
Sub Sector 206 d6
Sub Sector 207 d7
Sub Sector 208 d8
Sub Sector 209 d9
Year 1986 cp 1
Year 1987 (p 2
Year 1988 (p 3
Year 1989 (p 4
Year 1990 (p 5


in the fixed effects model for the FPB


Estimate
0.060
0.046
0.070
0.092
-0.025
0.057
0.056
0.012
0.046
0.061
0.098
0.063
0.059
0.060


Standard Error
0.030
0.027
0.017
0.018
0.022
0.019
0.031
0.019
0.020
0.019
0.019
0.019
0.019
0.019


t Value
2.010
1.680
4.040
5.130
-1.130
2.960
1.820
0.620
2.260
3.200
5.170
3.380
3.040
3.170


Pr > Itl
0.046
0.094
<.0001
<.0001
0.261
0.003
0.070
0.539
0.025
0.002
<.0001
0.001
0.003
0.002


Label


Cross
Cross
Cross
Cross
Cross
Cross
Cross
Cross
Cross


Sectional
Sectional
Sectional
Sectional
Sectional
Sectional
Sectional
Sectional
Sectional


Effect
Effect
Effect
Effect
Effect
Effect
Effect
Effect
Effect


Time Series Effect
Time Series Effect
Time Series Effect
Time Series Effect
Time Series Effect









Abstract of Thesis Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Master of Science

MEASURING THE PROFITABILITY OF THE UNITED STATES FOOD SUPPLY
CHAIN: CROSS-SECTION AND TIME-SERIES EFFECTS

By

Youshan Zhao

August 2010

Chair: Richard N. Weldon
Major: Food and Resource Economics

Over time, agricultural producers and agribusinesses in the U.S. food supply chain

have experienced a lot of market changes and restructuring. Agribusiness

management is increasingly focusing on the strategy to improve its profitability. There

are many key drivers to improve agribusiness's competitiveness and financial

performance, such as market demand, technical efficiency, environmental impact, etc.

This study concentrates on the operating strategy employed by the U.S. food supply

sectors by analyzing groups of financial ratios. Following the framework of the DuPont

system, this research classifies financial ratios into three categories to measure

agribusiness's operating efficiency, asset use efficiency, and capital structure.

Moreover, both fixed cross-section effect and time-series effect are examined though

the financial statement analysis with respect to a 23 years panel data. This research

aims to: 1) evaluate the critical operating concern for profitability of agribusiness; 2)

compare the profitability of sub sectors in the food supply chain; and 3) estimate the

time effect of economic recessions on agribusiness's profit rate.









financial ratios. It might be attributed to the relative small business scale in FWRS.

Table 3-1 shows that there are, on average, 125 firms in ten sub sectors of FPB, while

there are139 firms in the three sub sectors of FWRS. Especially, the food services sub

sector (SIC 581) has, on average, 89 firms per year. Many of those restaurants and

food providers are small-scale enterprises, which might be managed differently than

large firms. The high variability in production planning and action strategies for these

small businesses makes it difficult to predict their profitability using a specific model.

The change in EBTOCE is not majorly reflected by their financial ratios.

Regression results for testing the cross-section and time effects

The two-way fixed effect panel model is constructed to test hypotheses 2 and 3 at

the same time.

Test hypothesis 2: EBTOCE does not have cross-section effects within the U.S.

food supply chain. All sub sectors' profitability levels have no significant difference

statistically.

Test hypothesis 3: Sub sector's EBTOCE does not vary across time. There are no

uniform time effects to those sub sectors.

The null hypothesis is that parameters of sub sectors and time dummies are zero.

It is a twofold test. The first test is the cross sectional effect test. It captures the sub

sector effects which are constant over time. In this study, this refers to the within

differences of the component sub sectors in the FPB and FWRS. The second test aims

to capture the differences over 23 years that is common to all component sub sectors in

the U.S. food supply chain and to explore whether the time effect pattern follows the

recent three economic recessions. As stated in the Chapter 2, to avoid the perfect









different acceptable risk level for leverage. The interest expense might become a

burden for an overleveraged firm. Many firms find it difficult to repay their fixed interest

expense, so that they have to refinance old debt with new debt. This usually increases

their interest burden. It is a crucial management strategy to choose a debt and equity

mix to maximize profits. In the traditional DuPont system, leverage is measured by

assets divided by equity. From the basic accounting equation, assets=equity+ debt,

leverage rate can be transformed into total debt/ total assets (D/A).

We use this transformation because the financial debt ratios that involve equity in

denominator are unstable. Firms sometimes experience drastic net income fluctuations

which create problems in the calculation of equity financial ratios. For example, a

current period large net loss might result in two situations shown in the financial

statements. In the first situation, the net loss goes to the firm's retained earnings and

ends up with a negative equity balance. In the second case, after the firm's equity

absorbs the dramatic net loss in the current period, the equity turns to be a small

positive number. Thus, all financial ratios involving equity (from the balance sheet)

become incomparable with prior years. Besides, the change of debt structure has not

been measured consistently by this ratio due to the considerable change in equity.

Comparatively, using the total assets as denominator to measure the debt and equity

mix in a firm has less numerical disturbances. Given the robustness of total assets, the

equity is substituted by total assets for calculation of all debt ratios.

D/A is decomposed into the following function:

D/A=f (STD/A, CD/A, LTD/A, E), (2-3)









Table 4-3. Hypothesis test for fixed cross-section effects and time effects in the FPB
Panel A: Model Description
Estimation Method Fix Two
Number of Cross Sections 10
Time Series Length 23
Table 4-3 provides the regression results of cross-section effects and time effects in model: EBTOCE(it) = a0+31
GM/S(it)+ 32 SG&A/S(it)+ 33 IN/S(it)+ 34R&D/S(it)+ 35 ARTR(it)+ 361NTR(it) + 37PP&ETR(it)+ 38STD/A(it)+ 39CD/A(it)
+ 310OLTD/A(it)+ ui+ vt +E(it).









formula but are proxies of micro aspects of firm's operating condition. There are

numerous ratios that can be used to make comparisons across firms. It is feasible to

adopt any measurement with respect to the return on capital as long as this benchmark

is used consistently across all the samples. The above ratio substitutions are computed

over all firms consistently. Variable definitions and associated formulas are presented in

Table 2-2.

Statistical Model

Basic Regression Model

To test the linear relationship of the sub component ratios and the firm's total

profitability, those ten underlying ratios are treated as independent variables. The firm's

operating condition is reflected in the financial ratios in all aspects (operating efficiency,

asset use efficiency and financial leverage). Consequently, the ten financial ratios

contribute to the total profitability. The following regression is used as foundation of

later cross-sectional and time effect analysis:

Yit =ao+p' Xit + :it. (2-4)

EBTOCEit = ao+31 GM/Sit+ 32 SG&A/Sit+ 33 IN/Sit+ 34R&D/Sit+ 35 ARTRit

+ P6/NTRit + P7PP&ETRit+ 38STD/Ait+ 39CD/Ait

+ 31 OLTD/Ait + Eit. (2-5)

Where:

Yit = EBTOCE of the industry i in the year t,

ao = Intercept coefficient,

3' = Slope coefficient of regressors,

Xit = Independent financial ratios of industry i in the year t, and

Eit = Residual error for industry i in year t.









CHAPTER 1
INTRODUCTION

Profitability is defined as the ability for a firm to generate profits on a consistent

basis. In the modern business world, sustaining the profitability becomes a critical

success factor that leads to firm survival. The empirical tests in this study are based on

the profitability measurement by decomposing critical management factors.

Agribusiness Profitability

In the current dynamic and complex environment, having a good product or

service to customer does not guarantee high profitability. Profit is the difference

between total revenue and explicit costs. Numerous factors inside and outside the firm

could affect the agribusiness profitability. The internal factors that are affecting firm

profitability are expressed by management strategies such as business organization,

operating management, financial structure, etc. In contrast to those micro decisions,

the primary external market factors in determining firm success are macroeconomic in

nature, such as fiscal policy, governmental regulation, market competition, and product

demand, among others. A number of studies have explored the major determinants of

firm's profitability.

The agricultural sector in the United States is important, not only because the

agricultural products feed people in this country, but also because it is a supporting

industry to the economy of the United States. With the growing concentration of

ownership and centralization in food sectors, the agribusiness is the second most

profitable industry in the United States, following pharmaceuticals (Magdoff, Foster, and

Buttel, 2000). More and more research is expressing interest in the analysis of the









Among all sub sectors, tobacco sub sector and beverage sub sector have the

highest gross margin rate of 51.73% and 51.04% respectively. The bakery and

beverage sub sector spends the highest amount on interest, selling and G&A per dollar

of sales. The tobacco sub sector spends the highest amount on R&D per dollar of sales.

The food store and retailer have the best accounts receivable control. The food

services sub sector and food wholesaler enjoy the highest inventory turnover rate and

highest PP&E turnover rate respectively. The canned, frozen, and preserved fruits,

vegetables, and food specialties sub sector employs highest scale of short-term debt

comparing with other sub sectors. The top three sub sectors with the highest long-term

debt rates are retailer, tobacco, and food services. They all have a LTD/A of

approximately 30%.

One would expect that an economic recession would negatively affect the

EBTOCE of industries. However, the EBTOCE fluctuations in figures 3-3 and 3-4

indicate that this is not necessarily true in the food supply chain. The figures show a lot

of variation of EBTOCE among the different sub sectors, some of which can be

explained by economic booms and busts, but there are other factors that need to be

considered, like operating strategies.

In Figure 3-3, EBTOCE changes in food wholesale, retail, and services show that,

in the early 1990s recession, only the food wholesale sub sector (SIC 514) exhibited a

sharp decrease in returns. In the recession of 2007-2008, all industries except the food

wholesale sub sector experienced a fall in their EBTOCE. For the early 2000s

recession, it appears that recession did not have much impact on EBTOCE for the three

sub sectors in FWRS. Clearly, other factors are more important in determining variation









Decomposition of Profit Margin

The PM measures firm's operating efficiency. As an important profitability ratio, the

net profit margin is an indicator of a company's pricing policies and its ability to control

costs. The DuPont system uses net profit after taxes as net income (NI). In this study,

the earnings before tax (EBT) is used as a proxy for NI. Thus the tax burden impact is

excluded. EBT is:

EBT = Net sales Cost of goods sold (COGS) -Selling, general and administrative

expenses- Interest expenses- Research and development expenses -Other

miscellaneous expenses.

Electing the most important components of EBT, four underlying financial ratios

are employed as determinants of PM. The function for profit margin takes the following

form:

PM = f (GM/S, SG&A/S, IN/S, R&D/S, E), (2-1)

where GM/S or gross margin rate = (net sales-COGS)/ (net sales); SG&A/S or selling,

general and administrative expense rate = (selling, general and administrative expense)

/ (net sales)2; IN/S or interest expenses rate = (interest expenses)/ (net sales); R&D/S

or research and development expense rate = (research and development expense)/

(net sales); and E refers to residuals.

Decomposition of Assets Turnover

The second part of the expansion, net sales over average total assets, refers to

ATO which measures efficiency of a company's use of its assets (funded by owners and



2 Selling expense is combined with general and administrative expense as SG&A/S in this study for the
reason that they share the same nature and provide information on whether the management is spending
efficiently or wasting cash flow.









MEASURING THE PROFITABILITY OF THE UNITED STATES FOOD SUPPLY
CHAIN: CROSS-SECTION AND TIME-SERIES EFFECTS















By

YOUSHAN ZHAO


A THESIS PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE
MASTER OF SCIENCE

UNIVERSITY OF FLORIDA

2010









utilization, household incomes, business profits and inflation all fall, while bankruptcies

and unemployment rates soar.

From the supply side, a recession causes firm output and sales to fall dramatically.

In a competitive market like the U.S., the difference in profitability among firms rises

sharply and the weak firms are forced to shut down. With fixed production costs and

lower sales, average costs of production rise during a recession. Lower sales and

higher costs squeeze profit margins, forcing companies to cut costs where they can.

This means purchasing less, closing down inefficient firms, reducing number of stores,

and laying off workers. As workers are laid off, unemployment increases, consumer

spending and consumer confidence crumbles, which reinforces the recession, and

makes it worse.

Research Objectives

There are three main objectives in this research. First, this research aims to

explore the variation of agribusiness profitability over a 23 years period and identify

variances that influence profitability. To address this objective, I will use the extended

DuPont formula as the framework to analyze the financial components that affect firm

profits. The extended DuPont has ten sub financial ratios that will help determine

agribusiness' profitability.

The above statement leads to the first hypothesis;

Hypothesis 1

The sub component financial ratios of the DuPont expansion are not jointly

significant to explain the variation of profitability, and each independent financial ratio

has no effect on the profitability among all sub sectors of food supply chain.









in EBTOCE such as debt pressure, acquisition, cost control, exchange rate, etc. Taking

the food wholesale sub sector (SIC 514) as an example, this sub sector has the highest

return rate and volatility in its industry. The average EBTOCE for wholesale sub sector

is 13.85% with a standard deviation rate of 7.64% over past 23 years. The wholesale

sub sector experienced a trough in 1990. Seven out of eighteen wholesalers in that sub

sector had negative returns. Those seven firms were all middle to small size

wholesalers, and all had negative margin ratios along with increased financial leverage.

Some middle size wholesalers like Balfour Maclaine have exited the market since 1990.

From the year 1998, the largest wholesaler Fleming Companies started shrinking their

market shares due to three consecutive net operating losses. Increased competition

and failure to achieve necessary cost savings were mentioned as causes of net

operating losses. In 1998, its EBTOCE was as low as -24.00%, which pulled down the

weighted average EBTOCE of the wholesale sub sector.

By contrast, another major market player SYSCO CORP was successful in

increasing their market share in sales. SYSCO CORP started dominating the wholesale

industry from 1999 by increasing its return rate steadily from 22% to over 30 percent.

The strong growth of EBTOCE for the whole sub sector from year 2002 can be

attributed to the high return rate of SYSCO CORP.

The meat sub sector (SIC 201), as another example, experienced a huge

decrease in return starting in 2007. However, this decrease cannot be attributed to the

late 2000's recession, but rather to a bad acquisition and a huge debt load held by

Pilgrim's Pride, which was the largest chicken producer in the U.S. The Pilgrim's Pride

acquisition of rival Gold Kist for $1.3 billion in 2006 is the main reason for the company's









performance of agricultural and food system in terms of profitability (Kennedy, Harrison,

and Piedra, 1998; Neibergs, 1998).

Agribusiness and the U.S. Food Supply Chain

Agribusiness has long been recognized as an important industry and supporting

component in the U.S. economy. Being the world's leading exporter of agricultural

products, the U.S. agribusinesses supply food to the world. The U.S. is the largest

agricultural export country and has a surplus in agriculture despite of overall trade

deficit. According to the yearly trade data released by the USDA's Economic Research

Service and Foreign Agricultural Service, the U.S. exports agricultural products at a

value of $98,611 million, which is 11% of its total export goods to the world in 20091.

The agricultural sector also plays an important role in the U.S. employment.

Agribusiness spans a wide range of sectors in the U.S. food system. "The food

supply chain is a series of links and inter-dependencies, from farms to food consumers'

plates, embracing a wide range of disciplines (Bourlakis and Weightman, 2004)."

Generally, the food supply chain is recognized as the value chain in the agribusiness

system.

Economic Recession

A recession is defined as a decline in the Gross Domestic Product (GDP) for two

or more consecutive quarters. In a recession, business activity slows down,

unemployment increases, and the standard of living diminishes. A drop in consumer




1 Available at USDA's Economic Research Service website: Foreign Agricultural Trade of the United
States (FATUS) Data Sets 2010,http://www.ers.usda.gov/Data/FATUS/. Accessed on July 2010.









match with the three economic recessions during the observation range. Corresponding

to the fluctuation analysis of data in the Chapter 3, the change of EBTOCE attributes to

the management strategy of agribusiness rather than economic downturns. Generally,

the overall demand shrinks along with the recession. However, it seems to have no

statistically significant impact to the FPB industry's profitability, in terms of EBTOCE.

The two-way fixed-effects model output is provided in Table 4-5 for FWRS

industry. Panel A shows that there are three cross sections and 23 time observations.

Three cross sections are sub sector SIC 54, SIC 514 and SIC 581. Identical with

previous definition, 23 time observations cover the periods from 1986 to 2008. The F

test for fixed effects is shown in the Table 4-5 Panel B provides evidence that there are

no fixed effects. With the F-statistic of 1.27 and associated P value of 0.259, the null

hypothesis is not rejected. There are no sub sector effects, or time effects. To explore

the details of fixed effects, two separate one-way fixed effect regression models are

developed regarding the sub sector difference and time impact respectively. Table 4-5

Panel C reports two separate F-test. Both the p value for the F-Statistic is above 10%

significant level, which indicates an absence of any sub sector effect and time effects.

The sub sector dummies and year dummies are not statistically significant in this

industry. In essence, there is no statistical difference in EBTOCE among sub sectors in

FWRS over the past 23 years. The statistic output of FWRS in the Table 4-6

corroborates this conclusion. Twenty of the P values of coefficient estimates are greater

than 5%. That result might be attributed to the fact that there are too many small-scale

agribusinesses in FPB groups, like groceries and restaurants. They are more variable

and complex in profitability than large companies. There might be no common operating









CHAPTER 4
EMPIRICAL RESULTS

Before performing the regression, the regression of residuals for the fixed effect

panel model is tested. Without the obvious heteroskedastic error and autocorrelation,

the model exhibits an overall linear good fit.

Regression results for testing the model

After checking the residuals of the model, hypothesis 1 proposed in the Chapter 1

is tested.

Test hypothesis 1- the component financial ratios of the DuPont expansion have

the same effects on the profitability, in terms of EBTOCE, among all sub sectors that

comprise of the food supply chain.

Considering the different management strategies will be adapted to different

industries, regression tests are performed separately for the FPB and the FWRS, which

comprise ten and three sub sectors respectively.

As stated in the Chapter 2, each DuPont component financial ratio is measuring

one proxy variable of sub sector's performance. The two-way fixed effects panel

regression is based on the weighted average mean using net sales as the weight. The

empirical results are reported in Tables 4-1 and 4-2 for the two main industries. Since

this research is carried out in two stages, the cross-section and time effects results will

be shown later as part of the testing of hypotheses 2 and 3.

The F-test for the two-way fixed effects model focuses only on the presence of the

fixed effects, not on the significance of the explanatory variables. Thus, the OLS

regression was run to obtain the F-statistics for the ten component financial ratios










Table 4-2. Two-way fixed effects panel regression in FWRS industry
Model Description and Estimation
Total sample 69 Degree of freedom 33
Fit Statistics (F-statistics 9.77; p-value: 0.00)
R-Square 0.7955 MSE 0.0011
Parameter Estimates
Estimate Standard Error t Value Pr > Itl
Intercept -0.141 0.211 -0.670 0.510
GM/S 1.337 0.846 1.580 0.124
IN/S -4.460 1.378 -3.240 0.003
SG&A/S -0.600 1.119 -0.540 0.596
R&D/S
ARTR 0.000 0.001 -0.350 0.728
INTR 0.008 0.005 1.500 0.142
PP&ETR 0.001 0.007 0.200 0.843
STD/A -1.824 0.970 -1.880 0.069
CD/A -0.038 0.370 -0.100 0.919
LTD/A -0.558 0.138 -4.050 0.000
Table 4-2 presents the statistical output of the regression for the independent variables in FWRS INDUSTRY. Since the hypothesis
1, 2, and 3 are tested separately, the two-way fixed effects F test and results are moved to Table 4-3. The regression takes the form
that EBTOCE(it) = a(i)+31 GM/S(it)+ 32 SG&A/S(it)+ 33 IN/S(it)+ 34R&D/S(it)+ 35 ARTR(it)+ 361NTR(it) + 37PP&ETR(it)+
38STD/A(it)+ 39CD/A(it) + 310LTD/A(it)+ ui+ vt +E(it). Independent variables are defined as following: GM/S is gross margin rate;
SG&A/S is selling and general and administrative Expense rate; IN/S is Interest Expenses rate; R&D/S is research and development
expense rate; ARTR is accounts receivables turnover rate; INTR is inventory turnover rate; PP&ETR is property, plant and
equipment turnover rate; STD/A, CD/A and LTD/A represent short term debt, current portion of long-term debt and long-term debt to
assets ratio respectively.
R&D/S coefficient estimate and standard error are 0 because there is no R&D expense in FWRS industry.









parameter estimate of sub sector tobacco (SIC 21) (d10) in year 2008 ((p 23) is the

intercept (-0.038), a reference point. The other dummy parameter coefficients are

computed using this reference point as follows: intercept+ sub sector dummy coefficient

+ year dummy coefficient+ other dummy coefficients*0. The actual intercept of meat sub

sector (dl) in year 2007 ((p 22), for example, is computed as intercept+dl coefficient + cp

22 coefficient +other dummy estimates*0. That is 0.043=-0.038+0.06+0.021+0. The

coefficient 0.043 means, holding other explanatory variables constant, the EBTOCE of

meat sub sector in 2007 is 4.3% higher than tobacco sub sector in 2008. The

coefficients of other sub sectors years can be estimated similarly.

Looking at the parameters in Table 4-4, most of the cross-section effects are

highly significant (with the exception of sub sectors 202, 205, 207 and 208). Tobacco

sub sector (SIC 21) as a controlling group is sorted and dropped by SAS as a default.

The small P value of F-statistics means that the other sub sectors are significantly

different from the tobacco sub sector. A significant influence of sub sector-specific

factors is present.

Even thought in 12 out of 23 time periods the time effect is statistically significant,

the time series effects have no uniform pattern in FPB. It is shown that there are some

significant fluctuations occurred in the past 23 years, but not all sub sectors experienced

the significant decrease in return in a same certain year. Holding other independent

variables constant, the profitability of a sub sector in a specific recession period does

not exhibit the significant decrease in return comparing with other time periods. From

year 2000, however, the year dummy variables lose their significance. The parameters

values decrease in size. The interesting fact is, the visible time effects pattern do not









debt load. In 2008, the Pilgrim's Pride filed for bankruptcy, due to debt and high

commodity prices of feed inputs.

Those examples of the agribusiness' performance indicate that the return on

capital employed is more sensitive to management's strategies than to business cycles.

The cross section and time effects to the food supply sectors will be tested in Chapter 4.









Table 4-5. Continued
Panel B : Fixed effect F test
Model Fixed effect Sample Size F Test Prob value Result
Two-way Both sectors and time: 69 F(24,33)= 1.27 0.2588 Null hypothesis of
I =54,514,581 both sector and time
T =1986, 1987, 1988.....,2008 effect are not rejected












0.35

0.30

0.25 0

0.20 -

0.15 N

0.10

0.05

0.00

-0.051986

-0.10

-0.15


2004 2006


- SIC 201 MEAT SIC 202 Diary SIC203C.F, and PF and V


-SIC 204 Grain mill


-SIC 205 Bakery


Figure 3-4. EBTOCE fluctuation by weighted average in the FPB industry


1988 1990 1992 1994 V 1996 1998 2000 2002









model. It aims to evaluate the null hypothesis that all regression coefficients are equal to

zero versus the alternative that at least one is not.

The results for the food processing and beverage (FPB) industry are shown in

Table 4-1. The two-way fixed effects regression model for the FPB has an R-squared of

0.7182. F-statistics for the independent variables in the model is 19.23, with a p-value of

0.000. This indicates that 71.82% of a change in EBTOCE can be explained by the

changes of independent regressors. In other words, the ten sub components financial

ratios in this model explain 71.81% of EBTOCE's fluctuation in their linear relationship.

The F test from the OLS regression shows the usefulness of this model. The proposed

relationship between the EBTOCE and the set of financial ratios is statistically reliable.

The F test shows overall significance of this model. However, not all the ratios are

statistically significant. GM/S, IN/S, SG&A/S and R&D/S, proxies of profit margin rate,

are all statistically significant at 5% level. In the FPB industry, EBTOCE is very

sensitive to the change of profit margin, which is reflected by a coefficient of 0.798. This

means, one percent increase in the gross margin rate will increase the EBTOCE by

0.798 percent. Similarly, one percent increase in IN/S, SG&A/S and R&D/S will

decrease the firm's EBTOCE by 1.488, 0.581 and 2.329 percent, respectively. It is

worth mentioning that the research and development expense in agribusiness has a

negative impact in its profitability. R&D expenses in the food processing and beverage

sector have not been effectively increasing contemporaneous profitability. That might be

attributed to the lagged effect of research and development expenditures.

The second component tested is the statistical significance of the asset turnover.

Under this, there are three ratios- ARTR, INTR and PP&ETR. Out of these three, only









Table 4-6. Parameter estimates in the fixed effects model for FWRS industry


Parameter Estimates
Variable


Sub Sector
Sub Sector
Year
Year
Year
Year
Year
Year
Year
Year
Year
Year
Year
Year
Year
Year
Year
Year
Year
Year
Year
Year
Year
Year


54
514
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007


Sector 581 in year 2008


Variable Estimate
dl 0.151
d2 0.175
(p 1 0.092
p 2 0.116
p 3 0.129
p 4 0.153
p 5 0.179
p 6 0.123
(p 7 0.083
(p 8 0.061
(p 9 0.052
(p 10 0.030
p 11 0.024
(p 12 0.009
(p 13 0.015
(p 14 0.009
(p 15 0.008
(p 16 0.020
(p 17 0.023
(p 18 0.001
p 19 -0.018
p 20 -0.010
(p 21 -0.012
p 22 0.011
d3 p 22 -0.141


Standard Error
0.176
0.106
0.067
0.067
0.050
0.052
0.056
0.047
0.039
0.039
0.043
0.036
0.036
0.032
0.034
0.031
0.031
0.028
0.028
0.029
0.029
0.030
0.030
0.027
0.211


t Value
0.860
1.650
1.360
1.740
2.590
2.970
3.200
2.650
2.130
1.550
1.220
0.840
0.660
0.270
0.430
0.310
0.270
0.730
0.830
0.030
-0.630
-0.320
-0.410
0.420
-0.670


Pr > Itl
0.398
0.109
0.182
0.090
0.014
0.006
0.003
0.012
0.041
0.132
0.232
0.407
0.511
0.789
0.670
0.762
0.790
0.468
0.410
0.976
0.533
0.749
0.683
0.680
0.510


Label
Cross Sectional Effect
Cross Sectional Effect
Time Series Effect 1
Time Series Effect 2
Time Series Effect 3
Time Series Effect 4
Time Series Effect 5
Time Series Effect 6
Time Series Effect 7
Time Series Effect 8
Time Series Effect 9
Time Series Effect 10
Time Series Effect 11
Time Series Effect 12
Time Series Effect 13
Time Series Effect 14
Time Series Effect 15
Time Series Effect 16
Time Series Effect 17
Time Series Effect 18
Time Series Effect 19
Time Series Effect 20
Time Series Effect 21
Time Series Effect 22
Intercept


Ten explanatory variables parameter estimates are shown in table4-2.
SIC groups classifications follow: meat (201); diary (202); canned, frozen, and preserved fruits and vegetables (203);
grain mill (204); bakery (SIC 205); sugar and confectionery (206); fats and oils (207); beverages (208);miscellaneous food
preparations and kindred (209); tobacco (21); food service (5810 and 5812); retailers (5400 and 5411); and wholesalers (5140, 5141,
and 5180).









4 EMPIRICAL RESULTS ............. .... .............. ............ ............... 51

Regression Results for Testing the Model ............... ... .... ... ........... ....... 51
Regression Results for Testing the Cross-section and Time Effects .................. 54

5 C O N C LU S IO N ...................... ................ .. .. ................ ............... 70

LIST OF REFERENCES ................................. .................... 73

B IO G RA P H ICA L S KETC H ...................... .. ............. .. ......................... ............... 74










































6









Table 4-3. Continued
Panel C: separate F-test for individual effect
Model Fixed effect Sample Size F Test Prob value Result
Null hypothesis for


Single sectors effect sector effect is not
One-way I =201, 202, 203..., 21 230 F(9,208)= 9.21 <.0001 rejected
Single time effect
T =1986, 1987, Null hypothesis for time
One-way 1988.....,2008 230 F(22, 195)=2.82 <.0001 effect is not rejected
Table4-3 provides the regression results of cross-section effects and time effects in model: EBTOCE(it) = a0+31
GM/S(it)+ 32 SG&A/S(it)+ 33 IN/S(it)+ 34R&D/S(it)+ 35 ARTR(it)+ 361NTR(it) + 37PP&ETR(it)+ 38STD/A(it)+ 39CD/A(it)
+ 310LTD/A(it)+ ui+ vt +E(it).




Full Text

PAGE 1

1 MEASURING THE PROFITABILITY OF THE UNITED STATES FOOD SUPPLY CHAIN: CROSS SECTION AND TIME SERIES EFFECTS By YOUSHAN ZHAO A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE MASTER OF SCIENCE UNIVERSITY OF FLORIDA 2010

PAGE 2

2 2010 Youshan Zhao

PAGE 3

3 To Jingqi Wang, Zhenya Zhao and Qipeng Zheng

PAGE 4

4 ACKNOWLEDGMENTS I extend my deepest gratitude to the chair of my supervisory committee, Dr. Richard Weldon. Special thanks should be given to my committee member Dr. Carlos J. Trejo Pech and Dr. Randle. I acknowledge and express my sincere appreciation for their directio n, assistance, guidance, and for their encouragement to successfully complete the Master of Science program.

PAGE 5

5 TABLE OF CONTENTS page ACKNOWLEDGMENTS .................................................................................................. 4 LIST OF TABLES ............................................................................................................ 7 LIST OF FIGURES .......................................................................................................... 8 ABSTRACT ..................................................................................................................... 9 CHAPTER 1 INTRODUCTION .................................................................................................... 11 Agribusiness Profitability ......................................................................................... 11 Agribusiness and t he U.S. Food Supply Chain ....................................................... 12 Economic R ecession .............................................................................................. 12 Research O bjectives ............................................................................................... 14 Test H ypothesis 1 ............................................................................................. 14 Test H ypothesis 2 ............................................................................................. 15 Test H ypothesis 3 ............................................................................................. 16 2 LITERATURE REVIEW AND RESEARCH METHODOLOGY ................................ 17 Literature R eview .................................................................................................... 17 Total O rofitability M easurement .............................................................................. 20 Basic DuPont F ormulation ................................................................................ 20 Expansion of DuPont Formula .......................................................................... 20 Decomposition of Profit Margin .................................................................. 21 Decomposition of Assets Turnover ............................................................ 21 Decomposition and Substitution of Leverage ............................................. 22 EBTOCE in P rofitability M easurement ....................................................... 25 Statistical M odel ...................................................................................................... 26 Basic R egression M odel ................................................................................... 26 Panel D ata ........................................................................................................ 27 3 DATA SOURCES AND DESCRIPTION .................................................................. 32 Data and Trends ..................................................................................................... 32 Data S ources .................................................................................................... 32 Firm N umber F luctuations ................................................................................ 33 Weighted A verage M ethod ..................................................................................... 34 Summa ry S tatistics and A ccounting V ariables ........................................................ 35

PAGE 6

6 4 EMPIRICAL RESULTS ........................................................................................... 51 Regressi on R esults for T esting t he M odel .............................................................. 51 Regression R esults for T esting the C ross section and T ime E f f ects ...................... 54 5 CONCLUSION ........................................................................................................ 70 LIST OF REFERENCES ............................................................................................... 73 BIOGRAPHICAL SKETCH ............................................................................................ 74

PAGE 7

7 LIST OF TABLES Table page 2 1 Sub components of profitability measurement .................................................... 30 2 2 Variable definition ............................................................................................... 31 3 1 Firm number change during the past three recessions ....................................... 41 3 2 EBTOCE of FWRS industry by year and sub sector ........................................... 42 3 3 EBTOCE of FPB industry by year and sub sector .............................................. 43 3 4 Summary statistics of explanatory variables ....................................................... 44 4 1 Two way fixed effects panel regression in FPB industry .................................... 59 4 2 Two way fixed effects panel regression in FWRS industry ................................. 60 4 3 Hypothesis test for fixed cross section effects and time effects in the FPB ........ 61 4 4 Parameter estimates in the fixed effects model for the FPB ............................... 64 4 5 Hypothesis test for the fixed cross section effects and time effects in the FWRS industry ................................................................................................... 66 4 6 Parameter estimates in the fixed effects model for FWR S industry .................... 69

PAGE 8

8 LIST OF FIGURES Figure page 3 1 Firm N umber in FPB I ndustry ............................................................................. 46 3 2 Firm N umber in the FWRS I ndustry .................................................................... 47 3 3 EBTOCE F luctuation by W eighted A verage in the FWRS I ndustry .................... 48 3 4 EBTOCE F luctuation by W eighted A verage in the FPB I ndustry ........................ 49

PAGE 9

9 Abstract of Thesis Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Science MEASURING THE PROFITABILITY OF THE UNITED STATES FOOD SUPPLY CHAIN: CROSS SECTION AND TIME SERIES EFFECTS By Youshan Zhao August 2010 Chair: Richard N. Weldon Major: Food and Resource Economics Over time, agricultural producers and agribusinesses in the U.S. food supply chain have experienced a lot of market changes and restructurings. Agribusiness management is increasingly focusing on the strategy to improve its profitability. There are many key drivers to improve agribusinesss competitiveness and financial performance, such as market demand, technical efficiency, environmental impact, etc. T his study concentrates on the operating strategy employed by the U.S. food supply sectors by analyzing groups of financial ratios. Following the framework of the DuPont system, this research classifies financial ratios into three categories to measure agr ibusinesss operating efficiency, asset use efficiency, and capital structure. Moreover, both fixed crosssection effect and time series effect are examined though the financial statement analysis with respect to a 23 years panel data. This research aims t o: 1) evaluate the critical operating concern for profitability of agribusiness; 2) compare the profitability of sub sectors in the food supply chain; and 3) estimate the time effect of economic recessions on agribusinesss profit rate.

PAGE 10

10 This study provides information regarding the financial performance of agribusiness. And it advances the understanding of operating efficiency and effectiveness of agribusiness.

PAGE 11

11 CHAPTER 1 INTRODUCTION Profitability is defined as the ability for a firm to generate profits on a consistent basis. In the modern business world, sustaining the profitability becomes a critical success factor that leads to f irm survival. The empirical tests in this study are based on the profitability measurement by decomposing critical management factors. Agribusiness Profitability In the current dynamic and complex environment, having a good product or service to custome r does not guarantee high profitability. Profit is the difference between total revenue and explicit costs Numerous factors inside and outside the firm could affect the agribusiness profitability. The internal factors that are affecting firm profitability are expressed by management strategies such as business organization, operating management, financial structure, etc. In contrast to those micro decisions, the primary external market factors in determining firm success are macroeconomic in nature, such as fiscal policy, governmental regulation, market competition, and product demand, among others. A number of studies have explored the major determinants of firms profitability. The agricultural sector in the United States is important, not only because the agricultural products feed people in this country, but also because it is a supporting industry to the economy of the United States With the growing concentration of ownership and centralization in food sectors, the agribusiness is the second most pr ofitable industry in the United States, following pharmaceuticals (Magdoff, Foster, and Buttel, 2000) More and more research is expressing interest in the analysis of the

PAGE 12

12 performance of agricultural and food system in terms of profitabilit y (Kennedy, Harrison, and Piedra, 1998 ; Neibergs, 1998) Agribusiness and the U.S. Food Supply Chain Agribusiness has long been recognized as an important industry and supporting component in the U.S. economy. Being the world's leading exporter of agricult ural products, the U.S. agribusinesses supply food to the world. The U.S. is the largest agricultural export country and has a surplus in agriculture despite of overall trade deficit. A ccording to the yearly t rade data released by the USDA's Economic Resea rch Service and Foreign Agricultural Service, the U.S. exports agricultur al products at a value of $ 98,611 million which is 11% of its total export goods to the world in 20091. The agricultural sector also plays an important role in the U.S. employment. Agribusiness spans a wide range of sectors in the U.S. food system The food supply chain is a series of links and inter dependencies, from farms to food consumers plates, embracing a wide range of disciplines (Bourlakis and Weightman, 2004). Generally the food supply chain is recognized as the value chain in the agribusiness system. Economic R ecession A recession is defined as a decline in the Gross Domestic Product (GDP) for two or more consecutive quarters. In a recession, business activity slows down, unemployment increases, and the standard of living diminishes. A drop in consumer 1 Available at USDA's Economic Research Service website: Foreign Agricultural Trade of the United States (FATUS) Data Sets 2010, http://www.ers.usda.gov/Data/FATUS/. Accessed on July 2010.

PAGE 13

13 spending is generally a sign of a recession, which is caused by a mismatch of aggregated demand and aggregate supply. The fluctuation of production or economic activi ty between periods is described as a business cycle. It has a sequence of four phases: peak, contraction, trough, and expansion. A peak is defined as the time period when business activity is at its maximum. However, this maximum point does not last forever, and business activity starts to decline and contract. A trough is the time period when business activity reaches a minimum. However, business activities start to pick up, this is the expansionary period. Since 1854, the U.S. has encountered 32 cycles of expansion and contraction. Significant amount of research has been done to explain the relationship of recession and aggregate demand changes. In this study, however, the impact of the recent three main recessions on the supply side is examined. The three major recessions considered in this research are: 1) July 1990March 1991: 8 months (the early 1990s recession), 2) March 2001November 2001: 8 months (the early 2000s recession), and 3) December 2007to date: more than 25 months (the late 2000s recessio n). From the demand side, a recession reduces consumers ability and confidence to spend. The consumers, who overspent based on credit secured by bubble priced assets, drove the U.S. economy. With the decline in the bubble assets, c onsumers willingness to spend was hit hard, shaking their spending confidence. Industry wa s adversely affected as sales that depend on consumers consumption f ell. The effect of a recession was catastrophic: GDP, employment, investment spending, capacity

PAGE 14

14 utilization, household i ncomes, business profits and inflation all fall, while bankruptcies and unemployment rates soar. From the supply side, a recession causes firm output and sales to fall dramatically. In a competitive market like the U S the difference in profitability among firms rises sharply and the weak firms are forced to shut down. With fixed production costs and lower sales, average costs of production rise during a recession. Lower sales and higher costs squeeze profit margins, forcing companies to cut costs wher e they can. This means purchasing less, closing down inefficient firms, reducing number of stores, and laying off workers. As workers are laid off, unemployment increases consumer spending and consumer confidence crumble s, which reinforces the recession, and makes it worse Research O bjectives There are three main objectives in this research. First, this research aims to explore the variation of agribusiness profitability over a 23 years period and identify variances that influence profitability. To addr ess this objective, I will use the extended DuPont formula as the framework to analyze the financial components that affect firm profits. The extended DuPont has ten sub financial ratios that will help determine agribusiness profitability. The above stat ement leads to the first hypothesis ; Hypothesis 1 The sub component financial ratios of the DuPont expansion are not jointly significant to explain the variation of profitability, and each independent financial ratio has no effect on the profitability among all sub sectors of food supply chain.

PAGE 15

15 The second objective of the study is to examine whether the agribusinesss profitability varies across sub sectors within the U.S. food supply chain. Industrial economists define an industry as a group of firms that produce homogenous products such that consumers consider them substitutes. With similar operating features, agribusinesses in the same sub sector (i.e., same group under 3 digit SIC) might have similar customer groups, production requirements, and or ganizational structure. From the financial viewpoint, those agribusinesses would tend to have comparable profit margins, asset turnovers and capital structures. The examination of the relationships between the profitability and cross section effect leads t o my second hypothesis: Hypothesis 2 The estimator s for the coefficients of crosssectional effects are not jointly significant. Profitability does not have cross sectional effects within the U.S. food supply chain. All sub sectors profitability levels have no significant difference statistically. The final objective of this research is to test whether there is a time fixed effect in the agribusinesss profitability. Are there time effects that are common to all sub sectors? For example, do all sub sector s experience a decrease in return in a specific recession? Some economists are of the view that only sales of luxury and nonessential goods will be significantly affected by recessions, while necessary goods like food, water, and clothing will not be affected as severely The food industry generally supplies necessities and food consumers are very reluctant to forgo consumption even during the hard times. While they might substitute margarine for butter, people will still need to satisfy their basic demands. This study explores whether the profitability fluctuates along with the economic downturns. Additionally, it evaluates the agribusinesses

PAGE 16

16 financial performance for subsectors of food supply chain by performing the fixed effect panel regression. The third hypothesis takes the following form: Hypothesis 3 The sub sectors profitability does not vary across time. There are no uniform time effects to those sub sectors. The rest of this paper is organized as follows. Chapter 2 reviews the literature, int roduces methodology and specifies the model. Chapter 3 summarizes the data selection criteria and provides descriptive statistics. In Chapter 4, hypotheses stated in Chapter 1 are tested. Empirical results are analyzed in Chapter 4 In Chapter 5, a discu ssion of the limitations of this study and direction for future research ends this thesis.

PAGE 17

17 CHAPTER 2 LITERATURE REVIEW AND RESEARCH METHODOLOGY This chapter will first provide the reader with a review of literature. The next section explores the DuPont decomposition and measurement proxies for the regression analysis Finally the foundation of the twoway fixed effect regression is introduced. Both tools will be used to derive the empirical results based on the agribusiness financial data. L iterature R eview Financial ratios analysis within an industry peer group has been traditionally viewed as a major method that compares and measures firms performance in the competitive market. A number of historical accounting ratios are viewed as information that can forecast performance. By measuring the typical values and patterns for ratios over time, financial statement analysis is presented as a matter of pro forma analysis of the future (N issim and P enman 2001) Among all the financial performance evaluation models the DuPont system is the most common used for providing fundamental analysis regarding the total profitability of the firm The interest in the DuPont system and profitability drivers analysis is developing the association between c urrent and future financial ratios. It is suggested that by widening the accounting information set, profitability measures such as return on equity ( ROE ) will exhibit greater predictive power. The firms time series behavior and difference in ratios ac ross firms at a given time can be reflected in the changes in earnings (Freeman and Penman, 1982) As two multiplicative components of DuPont analysis, profit margin ( PM ) and total asset turnover ( ATO ) have been extensively examined by decomposing from the return on total assets (ROA), which is a sub

PAGE 18

1 8 component of ROE In particular the common form to analyze ROA is performed by focusing on the return on net operating assets (RNOA). RNOA can be decomposed into PM and net operating assets turnover (NOATO)1. Previous research has found that there is a negative convex relationship between the level of PM and NO ATO. This property tends to cluster by industry and shows stronger when applied to industry averages (Selling and Stickney, 1989) In a view of cross section industries, pooled firms (regardless of the industries that the y belong to ) tend to achieve similar levels of RNOA by varying the combination of PM and NO ATO Furthermore, prior studies explore the interpretation power of the change s in PM and NO A TO metrics at the same time. Empirical results show that NO ATO has more persistent predictive power than PM. That is, the change in NO ATO explains a larger portion of the change in RNOA ( Fairfield and Yohn 2003 ). Nissim and Penman (2001) suggest a residual income valuation framework without considering the effects of financial leverage. It i s commonly believed that the firms choice of capital structure can be manipulat ed Relevant studies isolate the leverage and associated returns as discretional attributes by management but not a desirable operating profitability variable. However, the determination of firms leverage rate affects their residual returns. Jensen (1998) argues that the cost of leverage, both direct cost and indirect cost, might improve corporate performance by forcing managements to make valuemaximizing strategy to avoid debt pressure. For example, a firm with debt cost from borrowings might cut its underperforming production line more 1 N et operating assets are businesses operating assets minus its operating liabilities. It has been analyzed widely by separating operating activities from financing activities. The purpose of do ing that is to evaluate firms operating performance independently.

PAGE 19

19 readily than a firm that bears no fixed debt cost. In industry downturns, the highly leveraged firms lose substantial market share to their more conservatively financed competitors. There is a positive relationship between financial condition and firm performance ( O pler and S heridan 199 4) Considering the importance of leverage, this study will examine the agribusiness total profitability decomposing all three DuPont components (PM, ATO and Leverage) and their underlying ratios. The main drawback of the DuPont system, as claimed by B anker, Chang, and Majumdar (1993) is that this system and its underlying ratios provide only a gross aggregate measure for firms profitability. This system doesnt easily capture the microattributes of productivity of firms, like product mix, price rec overy, capacity utilization, etc. With respect to agricultural sectors, the economics literature provides multiple explanations for the profitability measures. Forster (1996) uses the capital asset pricing model (CAPM) to investigate the rate of return, solvency, liquidity and other financial ratio measurements in agribusinesses. Foster found that the capital structure and business risk of agribusiness are interrelated. Both of them ar e important determinants of stock returns. In a ddition to the specific within firm financial factors mentioned above, there is an increasing interest regarding agribusiness profitability in response to macroeconomic conditions. Empirical results indicate that macroeconomic conditions (i.e. fiscal policy, business cycles) have differing effects on agribusiness profitability dependent on a firms financial structure and their market segment (Neibergs, 1998) Regarding the agribusiness sector, the financial ratio analysis in this study follows from recognition of accounting ratios relationshi ps that determine total profitability.

PAGE 20

20 Besides, using the cross section time series data, we investigate how earnings components exhibit the timeseries properties and cr osssection properties. Total Profitability Measurement Basic DuPont F ormulation ROE measures the rate of return on the ownership interest in a firm. It is measured as net income (NI) divided by average total equity (ATE). The DuPont Company started usi ng the breakdown of ROE to measure efficiency at generating profits from every unit of shareholders' equity. Basically, the ROE can be decomposed into three parts. The following decomposition, also called the DuPont expansion is known as a strategic profi t model. Mathematically: (2 1) The first ratio in 2 1 refers to the PM; the second ratio refers to ATO, and the third ratio of the DuPont formula reflects the leverage of the firm. Multiplying the first two fractions (PM*ATO) will yield the return on assets (ROA), which provides a picture of the ability of using assets to earn a profit. Expansion of DuPont Formula Three ratios are expanded from the DuPont system to measure ROE ; PM, ATO and leverage. Those financial ratios will be decomposed to the next level as micro proxies of firm operation conditions. T he major variables substitutions are explained in the following sections.

PAGE 21

21 Decomposition of Profit Margin The PM measures firms operating efficiency. As an important profitabili ty ratio, the net profit margin is an indicator of a company's pricing policies and its ability to control costs. The DuPont system uses net profit after taxes as net income (NI). In this study, the earnings before tax (EBT) is used as a proxy for NI. Th u s the tax burden impact is excluded. EBT is : EBT = Net sales Cost of goods sold (COGS) Selling, g eneral and a dministrative expenses Interest expenses Research and development expenses O ther miscellaneous expenses. Electing the most important compon ents of EBT, four underlying financial ratios are employed as determinants of PM. The function for profit margin takes the following form : PM = f ( GM/S, SG&A/S, IN/S, R&D/S, ) (2 1 ) w here GM/S or gross margin rate = (net sales COGS)/ ( net sales ) ; SG&A/S or selling, general and administrative expense rate = ( selling, general and administrative expense) / ( net sales )2; IN/S or interest expenses rate = ( interest expenses ) / ( net sales ) ; R&D/S or research and development expense rate = ( research and development expense) / ( net sales ) ; and refers to residuals. Decomposition of A ssets T urnover The second part of the expansion, net sales over average total assets, refers to ATO which measures effi ciency of a company's use of its assets (funded by owners and 2 S elling expense is combined with general and administrative expense as SG&A/S in this study for the reason that t hey share the same nature and provide information on whether the managem ent is s pending efficiently or wasting cash flow

PAGE 22

22 borrowers) in generating sales revenue. For an agribusiness, the major asset items include accounts receivables ; inventory; property, plant and equipment. Thus, those asset turnover ratios play the major role in the determination of total asset turnover. A company may have other forms of assets, like cash and security investment, which are generally not as important as those mentioned. Those major asset items have more important commercial substance in explaining the assets usage performance. S imilar to PM, the following expression is used to estimate total asset turnover rate: AT O =f ( ARTR INTR, PP&ETR ) (2 2) where ARTR or a ccounts receivables turnover rate = ( net sales ) / ( average accounts receivables) ; INTR3 or inventory turnover rate = ( COGS ) / ( average inventory ) ; PP&ETR or property, plant and equipment turnover rate = ( n ets sales ) / ( average property, plant and equipment ) ; and Decomposition and Substitution of Leverage The third part of the DuPont formula is leverage, which is defined as total assets divided by average equity. Leverage reflects the financial structure of the company and it is highly influenced by the different business envir onments. Under economic downturn, firms might have to borrow more funds to keep operating Simultaneously, the financial institutions usually reduce credit which makes it even harder for firm to borrow money during downturns The leverage ratio will prov ide a good explanation on the influence of capital structure to a firms profitability Firms usually use debt to supplement owners investment and increase gains. However, these liabilities have fixed payment obligation on a regular basis (i.e. interest expense). Different firms have 3 Some practitioners prefer using net sales instead of COGS to calculate the INTR. This study adopts COGS because inventories are usually recorded at cost while net sales are recorded at market value.

PAGE 23

23 different acceptable risk level for leverage. The interest expense might become a burden for an overleveraged firm. Many firms find it difficult to repay their fixed interest expense, so that they have to refinance old debt with new debt. This usually increases their interest burden. It is a crucial management strategy to choose a debt and equity mix to maximize profits. In the traditional DuPont system, leverage is measured by assets divided by equity. From the basic acc ounting equation, assets=equity+ debt, leverage rate can be transform ed into total debt/ total assets (D/A). We use this transformation because the financial debt ratios that involve equity in denominator are unstable. Firms sometimes experience drastic net income fluctuations which create problems in the calculation of equity financial ratios. For example, a current period large net loss might result in two situations shown in the financial statements. In the first situation, the net loss goes to the f irms retained earnings and ends up with a negative equity balance. In the second case, after the firms equity absorbs the dramatic net loss in the current period, the equity turns to be a small positive number. Thus, all financial ratios involving equit y (from the balance sheet) become incomparable with prior years. Besides, t he change of debt structure has not been measured consistently by this ratio due to the considerable change in equity Comparatively, using the total assets as denominator to measur e the debt and equity mix in a firm has less numerical disturbances. Given the robustness of total assets, the equity is substituted by total assets for calculation of all debt ratios. D/A is decomposed into the following function: D/A=f ( STD/A, CD/A, LTD/A ) (2 3)

PAGE 24

24 where STD/A or Short term borrowings (including short term loan and short term note payable) to assets = (s hort term borrowings ) / (t otal assets ) ; CD/A or current porti on of long term debt to assets = ( current portion of debt ) / ( total assets ) ; Long term debt to assets ( LTD/A ) = ( long term debt ) / ( total assets ) ; Note that t he current portion of long term debt (CD) is presented separately with the short term debt in this study for the reason that these two types of debt have different interest rates and are represent ing different financing strategy of firm. The CD is also separated from the long term debt because it reflects the payment that is due within a year, which might be an element that pushes the management to tak e action to improve liquidity Many studies isolate the financial structure from the DuPont decomposition as a separate research field. T hey believe that unlike PM and ATO, leverage is a decision that is highly controlled by management. Some profitability valuation studies ignore the leverage factor. Authors believe the firms profitability is mainly driven by PM and ATO (Fairfield and Yohn 2001; Nissim and Penman 2001; Penman and Zhang 2003; Fairfield, Sweeney and Yohn). These studies argue that the financial structure and associated return should be ignored when performing profitability analysis, because it can be manipulated by management discretionally. In my study, I include the financial leverage ratio to measure the total profitability The financial structure captures the firms ability to use available economic resources to increase profit. This phenomenon is often referred to as the multiplier effect because the way of financing of the firm can affect the value of shareholders equity.

PAGE 25

25 The aforementioned ten determinants are disaggregated from the DuPont framework to measure agribusinesss profitability and perform further financial analysis ( Table 2 1). EBTOCE in P rofitability M easurement Similar to previous studies, average shareholders' equity is not a stable measurement due to the net earning fluctuation that can occur over time. In this study, the earnings before tax on capital employed (EBTOCE) is substituted for ROE. The calculation of EBTOCE is earnings before tax divided by capital employed. The EBTOCE is a relative more comprehensive measure of a firms ability to generate returns to pay for its cost of capital employ ed. Capital employed may be defined in a number of ways Here, I use the average net capital employed, which is the summation of fixed assets, investments, and net working capital. It represents the capital investment necessary for a business to operate. This substitution is made for three reasons. First, i nterest expense i s one of the research concerns in this study, which depicts the management financing strategy and affects agribusinesss performance. Whereas, as a fixed portion of net income, tax expense is excluded from the measurement of profitability. Second, the earnings in the numerator match the range of denominator: capital employed. Because the capital employed includes borrowings, so that the earnings should consider the interest cost to match the funds sources Third, a s long as EBTOCE is consistently computed for all firms, it has comparison power to interpret the profitability in terms of ratio EBTOCE. The overall approach of this study is based on the framework provided by the DuPont expansion. However, the financial ratios are not exactly derived from DuPont

PAGE 26

26 formula but are proxies of micro aspects of firms operating condition. There are numerous ratios that can be used to make c omparisons across firms It is feasible to adopt any m easurement with respect to the return on capital as long as this benchmark is used consistently across all the samples. The above ratio substitutions are computed over all firms consistently. Variable definitions and associated formulas are presented in Ta ble 2 2. Statistical M odel Basic Regression Model To test the linear relationship of the sub component ratios and the firms total profitability, those ten underlying ratios are treated as independent variables. The firms operating condition is reflected in the financial ratios in all aspects ( operating efficiency asset use efficiency and financial leverage). Consequently, the ten financial ratios contribute to the total profitability. The following regression is used as foundation of later cross sectio nal and time effect analysis: Yit = 0it + it (2 4) EBTOCEit 0 GM/Sit SG&A/Sit IN/Sit+ R& D/Sit ARTRit + INTRit + PP&ETRit+ STD/Ait+ CD/Ait + LTD/Ait + it. (2 5) Where: Yit = EBTOCE of the industry i in the year t 0 = Intercept coefficient Xit = Independent financial ratios of industry i in the year t and it = Residual error for industry i in year t

PAGE 27

27 Panel D ata Panel data sets usually include multiple cross sectional data (i.e. country, blocks, and firms) which are observed over two or more time periods. The characteristic of panel data was summarized by Yaffee, (2003) Panel data analysis endows regression analysis with both a spatial and temporal dimension. Panel data has become widely analyzed to understand social and economic changes over a time span. When both the space and time span is considered, the panel data analysis provides a number of ways to improve the interpretation of data. The OLS regression assumes that the residuals are independent. However, in a typical panel data set, the residual components in (2 5), are highly likely to be correlated both with the crosssection error and the time series erro rs. That violates the assumptions of OLS regression and can lead to biased estimates of coefficients and biased estimates of the standard errors. To avoid that common problem of the OLS model a two way fixed eff ects panel model is used in this research, which assumes that the error structure is corresponding to both cross section and time effect simultaneously. The fixed effects model is one of several types of panel analytic models. If the specification is depe ndent only on the cross section to which the observation belongs, such a model is referred to as a oneway model. When a data structure depends on both the cross sections and the time periods to which the observation belongs, it is called a twoway model. Additionally, if the data structure provides nonrandom effect pattern, it is called fixedeffects model. Otherwise it is the random effects models. The

PAGE 28

28 only difference is that random effects assume the intercept is uncorrelated with each explanatory variable. In this study, we assume that the agribusiness profitability is nonrandomly affected by both the cross section and timeseries. Because of the assumption that the crosssection and time series effects are fixed, the models are essentially regression models with dummy variables that correspond to the specified effects. This twoway fixed effects model is also refers to the twoway least square dummy variable model (two way LSDV). The two way fixed effects regression model applies to the DuPont compon ents are as follows: Yit = 0it + ui+ vt+ it (2 6) Where Yit = EBTOCE of the industry i in the year t 0 = Intercept coef ficient of dropped dummy industry i in the dropped year t Xit = Independent financial ratios of industry i in the year t ui = crosssection effects that are constant over time vt = time effects that are com mon to all groups and it = Residual error for industry i in year t This panel regression assumes that slopes are constant, only intercepts vary according to cross section and time. This model specifies i 1 sub sector dummies, t 1 time dummies to avoid perfect multicollinearity (the dummy variable trap). Multicollinearity refers to excessive correlation of the predictor variables. When

PAGE 29

29 correlation is excessive, the coefficients and standard errors of the independent variables become large, making it difficult or impossible to assess the relative importance of the predictor variables. Using the panel procedure, you do not need to create dummy variables and compute deviations from the group means. After sorting the cross section and year, it will drop the last sub sector and year dummy automatically. This procedure would repor t correct MSE, SEE, R2, and standard errors, and conducts the F test for the fixed group effect as well.

PAGE 30

30 Table 2 1. Sub components of profitability measurement Total profitability measurement Components ratios Profitability EBTOCE Operating efficiency GM/S SG&A/S IN/S R&D/S Asset use efficiency ARTR INTR PP&ETR Financial leverage STD/A CD/A LTD/A

PAGE 31

31 Table 2 2. Variable definition Abbreviation Description Formula ROE Return on equity N et income/ equity ROA Return on assets N et income/ total assets EBT Earnings before tax P retax income EBTOCE Return on capital employed E arnings before tax / capital employed PM Net profit margin rate P retax income/ net sales GM/S Gross margin rate ( N et sales COGS)/ net sales SG&A/S Selling,g eneral and administrative expense rate General and administrative expense / net sales IN/S Interest expenses rate Interest expenses/ net sales; R&D/S Research and development expense rate Research and development expense/ net sales. AT O Total assets turnover Net sales/ average total assets NOATO Net operating assets turnover Net sales/average net operating assets ARTR Accounts receivables turnover rate Net sales/ average accounts receivable INTR Inventory turnover rate COGS/ average inventory; PP& ETR Property, plant and equipment Nets sales / average property, plant and equipment. STD/A Short term debt to assets ratio Short term debt/ total Assets CD/A Current portion of long term debt to assets ratio Current portion of long term debt/ total assets LTD/A L ong term debt to assets ratio Long term debt/ total assets

PAGE 32

32 CHAPTER 3 DATA SOURCES AND DESCRIPTION Data and Trends Data Sources T he financial statement data for this study is derived from the CRSP/Compustat Merged Database (CCM), accessed from the Wharton Research Data Services (WRDS). The financial data of the U.S. agribusinesses food supply chain obtained covers 23 years from 1986 to 2008. During the 23 years, the U.S. went through three economic recessions: the early 1990s, the early 2000s, and the late 2000s recession. The major financial items used to calculate the financial ratios, the variables of main interest in this study, are drawn from the firms annual income statements and annual balance sheets. The Economic Research Service (ERS) is a primary source of economic information and research in the U.S. Department of Agriculture. The ERS classifies the farm and farm related industries based on Standard Industry Classification (SIC) codes. According to the ERS, there are six major industry groups which satisfy the demand for agricultural products. These six major groups are: Farm production, Agricultural services, forestry, and fishing Agricultural inputs industries Agricul tural processing and marketing Agricultural wholesale and retail trade, and Indirect agribusinesses.

PAGE 33

33 To be consistent with other food supply chain studies, the selection of agribusiness in this research is based on the 3 digit SIC code. Two major industries and their thirteen sub sectors are considered to comprise the U.S. food supply chain. One major industry is agricultural processing and marketing (referred to food processing and beverage, or FPB). This industry belongs to the manufacturing division. The second industry is agricultural wholesale and retail trade (referred to food wholesale, retail, and service, or FWRS). This group is classified as part of the wholesale trade and retail trade division. The ten s ub sectors in the FPB include: meat (SIC 201); dairy (SIC 202); canned, frozen, and preserved fruits, vegetables, and food specialties (SIC 203); grain mill (SIC 204); bakery (SIC 205); sugar and confectionery (SIC 206); fats and oils (SIC 207); beverages (SIC 208); miscellaneous food prepar ations and kindred (SIC 209); and tobacco (SIC 21). The three sub sectors of the FWRS are; wholesalers (SIC 514), retailers (SIC 54), and the food service industry (SIC 581). This classification follows Trejo Pech, Weldon, and House (2008). The sample contains 6157 firm year observations for the 19862008 time period. 88 firms had missing values for major financial items and were removed from the sample. The final sample includes 6069 firm year observations. 49.11% of the total observations belonged to the FPB, and 52.79% to the food wholesale, retail, and services. Firm Number Fluctuations Figures 31 and 32 show the change in the number of agribusinesses in the food supply chain over the past 23 years. The number of firms fluctuates within the years ac cording to the business peaks and troughs (business cycles). Table 3 1 summarizes

PAGE 34

34 the decrease of market players during economic downturns. For example, the average number of firm over the 23 years for the FPB is 125. The number of firms decreased by ten during the early 1990s recession (calculated as the average number of firms in the time period minus number of firms in 1990). Table 3 1 shows that the number of firms in each recession period is apparently lower than the average firm number over the past 23 years. Since it is difficult to get the firm number for the specific month during the recession, the number of firm number at the end of 1999 is used to represent the firm number in the early 1990s recession. Similarly, firm numbers at the end of 2001 and 2008 are used to represent the early 2000s recession and the late 2000s recession respectively. Weighted A verage M ethod In every stage of an industry, new firms enter while distressed firms leave. This causes the panel data to be variable and unbalanc ed. Additionally, particularly for the agribusiness sample, some firms operated for only a few periods during the past 23 years and then were merged or acquired. This is another reason for the unbalanced nature of the data set. Considering that the purpose of this study is to explore the sub sectors effects and time effect on the sub sectors EBTOCE as opposed to the individual firms, the sub sector average is obtained to perform the twoway fixed effect panel regression. The majority of past research employs the arithmetic mean of financial ratios as proxy of industry performance. But the straight average value does not reflect the industry objectively since it assumes that all firms have equal influence on that industry. I t is felt that a weighted mean o f financial ratios to measure the industrys primary financial condition is a better proxy of industry performance. In a competitive market,

PAGE 35

35 large firms tend to achieve economies of scale to gain competitive advantage. Emphasizing the resource efficiencies, productivity, and product quality, large firms dominate the development of industries. This can be justified because even though an industry may have many market players of all sizes, the market share is usually dominated by a single large firm or a few large firms. In my model, the net sales variable is applied as the base to calculate the weighted average. The final sample includes 6069 firm year observations After taking the weighted method to calculate the sector average, there are 299 (or 13*23) s ector year samples in the data set. Each variable in each sub sector is calculated as an average value yearly. By industries, there are 230 and 69 data lines in FRB and FWRS respectively after taking the weighted average by net sales. The unbalanced firm y ear panel data is transformed into a balanced sub sector year panel data. Tables 33 and 3 4 report the summary statistics based on weighted average method regarding the sub sectors over 23 years. Summary S tatistics and A ccounting V ariables Summary statis tics for all dependent and independent variables are provided in the Tables 32 to 34 based on 3 digit SIC sub sectors and their respective industry groups. Mean values are weighted by net sales. EBTOCE of the two industr ies are presented by sub sectors and year in the Table 3 2 and Table 3 3 respectively. For a better understanding of the EBTOCE fluctuation during the period of 19862008, F igure s 3 3 and 34 present the corresponding graphs for EBTOCE. Mean and standard dev iation of the ten explanatory ratios for all sub sectors in two industries of food supply chain are shown by pooled years in Table 3 4.

PAGE 36

36 The average profitability measured in terms of EBTOCE for the entire food wholesale, retail, and services over past 23 y ears was 11.39% with a standard deviation of 2.7 0 %. Breaking down the three sub sectors, the food wholesale sub sector ha s the highest profitability of 13.85 %, followed by food services with profitability of 10.5 % and finally the food retail sub sector w ith profitability of 9.81%. Their standard deviations have the same order with EBTOCE. The f ood wholesale sub sector is the most variable and food retail sub sector is relatively stable in return rate. For a better comparison, ten sub sectors w ithin the FP B industry are classified into two groups according to their SIC order. Each group presents 5 sub sectors. Figure34 show s yearly EBTOCE fluctuations for these ten sub sectors respectively. The EBTOCE in FPB ( Table 3 3) by year shows that all ten sub sectors have higher than 10 % profitability in terms of EBTOCE The average EBTOCE of the entire FPB during the past 23 years is 1 5.48% with a standard deviation of 2. 33 %. Average EBTOCE for the FPB is 4.09% higher than the FWRS Among the ten sub sectors of FPB sugar and confectionery is the most profitable sub sector with an average EBTOCE of 21.16%, followed by the canned, frozen, and preserved fruits, vegetables, and food specialties ( C F, and PF and V ) and grain mill with EBTOCE of 19.86% and 19.31% respectively The diary sub sector is more variable in profitability in terms of EBTOCE during the time period, shown by a standard deviation of 6. 83 %. Furthermore, all sub sectors except the tobacco and fats and oils sub sectors have relatively slight ly downward trend meaning that the average EBTOCE has been decreasing over the years. One possible explanation of falling profits is that both competition and technology have been increasing steadily which is shrinking profits.

PAGE 37

37 Table 3 4 provides the mean and standard deviation of the ten explanatory variables by all sub sectors among food supply chain over the 23 years Overall, in terms of gross margin rate, FPB industry outperforms the FWRS with 36.49% (standard deviation of 3.56%) compared to 22.08% (2.21%). The most profitable sub sectors in the food supply chain, like tobacco, s ugar and confectionery and beverage are all components of FPB. Besides, FPB operates more aggressively than FWRS by spending a larger portion of net sales on selling expense, G&A expense, interest expense, and R&D expenditure. From the perspective of assets use efficiency, companies with low profit margins usually tend to have high asset turnover This is true in food supply chain as well. FWRS has the higher assets turnover rate (means faster in times per year) than FPB. In comparison, the accounts receivable turnover rate in FWRS industry (33.89) is 2.97 times higher than FPB (11.43). That might be because the trading and services industry are dealing with final cus tomers so that it involves less receivables. The inventory turnover rate in FWRS (18.76) is 2.83 times faster than FPB (6.62), indicating that FWRS has better inventory management than manufactures in FPB. PP&E turnover rate in FWRS (7.83) is 1.86 times m ore than FPB (4.21), because the former industry requires less manufacturing equipments than the latter. The above indicates that the FWRS has higher assets usage efficiency in generating sales. With respect to capital structure, FPB has a similar portion of short term debt mixture but a slightly larger scale of long term debt mixture. It is worth to mention that long term debt is the main financing source for agribusiness. For the entire food supply chain, the short term debt (exclude current portion of lo ng term debt) is only 1.11%, compared with a long term debt to assets ratio of 27.32%.

PAGE 38

38 Among all sub sector s, tobacco sub sector and beverage sub sector have the highest gross margin rate of 51.73% and 51.04% respectively. The bakery and b everage sub sect or spends the highest amount on interest, selling and G&A per dollar of sales. The t obacco sub sector spends the highest amount on R&D per dollar of sales. The f ood store and retail er have the best accounts receivable control. The f ood services sub secto r and food wholesaler enjoy the highest inventory turnover rate and highest PP&E turnover rate respectively. The canned, frozen, and preserved fruits, veg etables, and food specialties sub sector employs highest scale of short term debt comparing with other sub sectors. The top three sub sectors with the highest long term debt rates are retailer, tobacco, and food services. They all have a LTD/A of approximately 30%. One would expect that an economic recession w ould negatively affect the EBTOCE of industries. However, the EBTOCE fluctuations in figures 3 3 and 34 indicate that this is not necessarily true in the food supply chain. The figures show a lot of variation of EBTOCE among the different sub sector s, some of which can be explained by ec onomic booms and busts, but there are other factors that need to be considered, like operating strategies In Figure 3 3, EBTOCE changes in food wholes ale, retail, and services show that, in the early 1990s recession, only the food wholesale sub sector (S IC 514) exhibited a sharp decrease in returns. In the recession of 20072008, all industries except the food wholesale sub sector experienced a fall in their EBTOCE. For the early 2000s recession, it appears that recession did not have much impact on EBTOC E for the three sub sector s in FWRS Clearly, other factors are more important in determining variation

PAGE 39

39 in EBTOCE such as debt pressure, acquisition, cost control, exchange rate, etc. Taking the food wholesale sub sector (SIC 514) as an example, this sub s ector has the highest return rate and volatility in its industry. The average EBTOCE for wholesale sub sector is 13.85% with a standard deviation rate of 7.64% over past 23 years. The w holesale sub sector experienced a trough in 1990. Seven out of eighteen wholesalers in that sub sector had negative returns. Those seven firms were all middle to small size wholesalers, and all had negative margin ratios along with increased financial leverage. Some middle size wholesalers like B alfour M aclaine have exited th e market since 1990. From the year 1998, the largest wholesaler F leming C ompanies started shrinking their market shares due to three consecutive net operating losses. Increased competition and failure to achieve necessary cost savings were mentioned as causes of net operating losses. In 1998, its EBTOCE was as low as 24.00 % which pulled down the weighted average EBTOCE of the wholesale sub sector By contrast, another major market player SYSCO CORP was successful in increasing their market share in sales. SYSCO CORP started dominating the wholesale industry from 1999 by increasing its return rate steadily from 22% to over 30 percent. The strong growth of EBTOCE for the whole sub sector from year 2002 can be attributed to the high return rate of SYSCO CORP. The meat sub sector (SIC 201), as another example, experienced a huge decrease in return starting in 2007. However, this decrease cannot be attributed to the late 2000s recession, but rather to a bad acquisition and a huge de bt load held by Pilgrim's Pride which wa s the largest chicken producer in the U.S. T he Pilgrim's Pride acquisition of rival Gold Kist for $1.3 billion in 2006 is the main reason for the company's

PAGE 40

40 debt load In 2008, the Pilgrim's Pride filed for bankrupt cy due to debt and high commodity prices of feed inputs Those examples of the agribusiness performance indicat e that the return on capital employed is more sensitive to managements strategies than to business cycles. The cross section and tim e effects to the food supply sect ors will be tested in Chapter 4.

PAGE 41

41 Table 3 1 Firm number change during the past three recessions Entire food supply chain FPB FWRS Firm number Mean Reduced Firm number Mean Reduced 1990 115 125 10 126 139 13 2001 115 125 10 122 139 17 2008 95 125 30 78 139 61

PAGE 42

42 Table 3 2 EBTOCE of FWRS industry by year and sub sector Sector Retailers Food wholesalers Food service SIC 54 514 581 Mean 1986 14.09% 12.45% 7.23% 11.26% 1987 16.82% 11.78% 11.01% 13.20% 1988 11.34% 9.94% 10.89% 10.72% 1989 10.27% 10.69% 9.11% 10.02% 1990 12.45% 0.50% 8.23% 6.73% 1991 12.38% 10.95% 9.26% 10.86% 1992 10.02% 10.55% 8.86% 9.81% 1993 10.27% 10.26% 3.86% 5.56% 1994 9.09% 10.90% 11.17% 10.39% 1995 12.03% 10.71% 8.35% 10.36% 1996 11.86% 9.45% 8.36% 9.89% 1997 11.48% 9.79% 7.36% 9.54% 1998 10.52% 0.76% 14.26% 8.51% 1999 11.49% 9.70% 11.94% 11.04% 2000 8.18% 7.48% 13.76% 9.81% 2001 8.28% 16.54% 13.08% 12.63% 2002 6.32% 24.43% 11.78% 14.18% 2003 3.97% 24.24% 11.14% 13.12% 2004 2.48% 25.08% 12.59% 13.38% 2005 6.66% 25.52% 14.12% 15.43% 2006 9.57% 20.41% 15.11% 15.03% 2007 10.89% 22.53% 15.27% 16.23% 2008 5.23% 24.96% 12.56% 14.25% MEAN 9.81% 13.85% 10.50% 11.39% Std 3.28% 7.64% 3.98% 2.70%

PAGE 43

43 Table 33. EBTOCE of FPB industry by year and sub sector SIC 201 202 203 204 205 206 207 208 209 21 Year Me at Diary C.F, and PF and V Grain mill Bakery Sugar and confectionery Fats and oils Beverages Miscellane ous food Tobacco Mean 1986 15.92% 21.87% 18.44% 24.97% 15.48% 18.34% 13.41% 14.65% 12.39% 21.30% 17.68% 1987 14.63% 19.76% 21.20% 29.64% 14.77% 42.88% 15.13% 17.23% 15.19% 24.76% 21.52% 1988 11.89% 24.10% 20.49% 25.19% 17.73% 24.13% 16.22% 18.76% 11.45% 15.02% 18.50% 1989 9.90% 20.30% 18.11% 22.04% 18.70% 24.58% 17.13% 17.52% 14.06% 13.84% 17.62% 1990 12.81% 19.40% 18.04% 24.54% 16.28% 24.78% 16.96% 18.10% 15.30% 15.40% 18.16% 1991 7.57% 16.56% 23.39% 24.98% 9.09% 21.26% 14.11% 17.98% 16.79% 15.67% 16.74% 1992 9.56% 13.41% 21.49% 23.87% 9.48% 19.95% 12.67% 18.86% 16.82% 18.69% 16.48% 1993 12.04% 13.36% 18.83% 21.84% 4.76% 22.43% 10.63% 19.62% 14.60% 13.80% 15.19% 1994 17.12% 12.12% 20.78% 21.49% 6.51% 21.32% 9.68% 20.76% 9.08% 17.79% 15.66% 1995 18.96% 4.75% 20.14% 30.72% 7.48% 21.23% 14.48% 19.94% 12.72% 20.13% 16.11% 1996 11.79% 12.05% 29.46% 21.24% 4.76% 21.90% 11.92% 18.98% 6.89% 23.47% 16.25% 1997 8.77% 10.29% 19.64% 8.82% 6.74% 28.18% 7.06% 21.66% 13.79% 22.26% 14.72% 1998 11.27% 5.23% 20.14% 20.05% 3.01% 20.34% 6.03% 18.50% 16.33% 18.18% 13.91% 1999 15.14% 9.06% 25.05% 24.36% 6.63% 25.71% 3.70% 16.47% 17.64% 21.72% 16.55% 2000 10.57% 8.14% 21.44% 24.92% 8.96% 19.69% 2.88% 15.21% 24.31% 19.45% 15.56% 2001 8.31% 4.08% 24.84% 9.51% 6.52% 19.83% 3.67% 18.75% 20.20% 16.92% 13.26% 2002 6.12% 8.68% 17.60% 11.84% 4.01% 21.61% 7.05% 18.71% 18.85% 19.28% 13.38% 2003 7.60% 9.90% 19.49% 12.55% 1.21% 15.87% 5.73% 16.32% 18.46% 12.78% 11.99% 2004 11.34% 7.07% 16.31% 13.78% 13.43% 17.04% 5.91% 17.68% 17.10% 15.84% 13.55% 2005 10.60% 6.01% 14.24% 10.59% 14.13% 17.12% 11.64% 17.16% 22.75% 16.72% 14.10% 2006 1.13% 9.73% 16.13% 11.05% 15.46% 16.89% 12.95% 16.81% 12.09% 17.55% 12.98% 2007 6.61% 7.52% 16.32% 13.56% 17.08% 12.46% 19.50% 18.62% 14.20% 20.06% 14.59% 2008 9.20% 5.08% 15.22% 12.51% 15.49% 9.20% 13.37% 9.55% 12.66% 31.43% 11.53% MEAN 10.02% 11.26% 19.86% 19.31% 10.34% 21.16% 10.95% 17.73% 15.38% 18.78% 15.48% Std 5.75% 6.76% 3.51% 6.83% 5.37% 6.36% 4.84% 2.42% 4.04% 4.20% 2.33%

PAGE 44

44 Table 34. Summary s tatistics of explanatory variables Explanatory Variables GM/S IN/S S,G&A/S R&D/S ARTR Industries SIC Mean Std Mean Std Mean Std Mean Std Mean Std Food processing and beverage: 36.49% 3.56% 2.04% 0.64% 22.40% 3.00% 0.41% 0.15% 11.43 1.49 Meat 201 11.93% 1.75% 0.98% 0.17% 6.50% 0.85% 0.03% 0.02% 18.93 2.29 Dia ry 202 26.28% 1.51% 1.47% 0.61% 18.11% 1.34% 0.01% 0.01% 12.22 0.95 C.F and PF and V 203 39.18% 3.12% 2.28% 0.44% 22.58% 2.24% 0.46% 0.06% 9.77 1.15 Grain mill 204 39.70% 10.36% 2.05% 0.49% 25.68% 8.69% 0.81% 0.15% 11.52 0.93 Bak ery 205 48.58% 3.56% 2.64% 2.21% 34.60% 5.61% 0.24% 0.35% 11.76 2.60 Sugar and confectionery 206 44.26% 4.51% 2.05% 0.51% 28.10% 2.15% 0.46% 0.22% 8.63 1.13 Fat s and oils 207 11.91% 2.57% 1.55% 0.55% 3.35% 0.80% 0.11% 0.04% 10.79 2.49 Beverages 208 51.04% 1.95% 2.66% 0.35% 31.47% 2.22% 0.25% 0.09% 9.30 1.05 Mis cellaneous food kindred 209 40.26% 2.73% 2.12% 0.36% 27.37% 1.92% 0.83% 0.28% 8.76 1.02 Tobacco 21 51.73% 3.53% 2.59% 0.69% 26.26% 4.21% 0.87% 0.32% 12.60 1.28 Food wholesale, retail, and services: 22.08% 2.21% 1.45% 0.47% 13.64% 1.47% 0.01% 0.01% 33.89 8.39 Food wholesalers 514 14.92% 2.86% 0.60% 0.19% 11.04% 2.00% 0.01% 0.02% 19.06 3.10 Foo d store retail 54 25.43% 1.61% 1.14% 0.21% 20.07% 1.12% 0.00% 0.00% 52.70 15.35 Food service 581 25.88% 2.17% 2.62% 1.01% 9.82% 1.30% 0.02% 0.01% 29.92 6.72 Total food supply chain 29.28% 2.88% 1.74% 0.55% 18.02% 2.24% 0.21% 0.08% 22.66 4.94

PAGE 45

45 Tabl e 34. Continued Explanatory Variables INTR PP&ETR STD/A CD/A LTD/A Industries SIC Mean Std Mean Std Mean Std Mean Std Mean Std Food proce ssing and beverage: 6.62 1.39 4.21 0.75 1.55% 2.40% 6.27% 3.16% 24.18% 5.97% Meat 201 11.25 2.66 6.67 1.43 1.63% 2.48% 4.39% 2.41% 26.08% 3.54% Dia ry 202 12.46 2.50 5.29 0.58 0.39% 0.68% 3.42% 2.24% 32.03% 13.60% C.F and PF and V 203 4.30 0.22 3.84 0.48 3.68% 3.80% 9.96% 4.18% 27.04% 8.11% Grain mill 204 6.30 0.56 3.79 0.88 2.61% 3.52% 9.70% 2.77% 25.31% 4.18% Bak ery 205 7.44 4.44 3.31 0.45 0.53% 0.82% 3.31% 3.77% 22.47% 11.09% Sugar and confectionery 206 4.59 0.38 3.54 0.34 0.59% 0.67% 9.20% 3.66% 16.52% 4.22% Fats and oils 207 7.39 1.26 4.50 1.91 0.55% 0.83% 4.20% 4.22% 19.81% 2.98% Beverages 208 5.57 0.83 2.74 0.20 1.44% 2.01% 6.85% 2.14% 24.33% 2.05% Mis cellaneous food kindred 209 4.03 0.51 4.11 0.72 1.41% 3.32% 6.99% 4.18% 22.37% 3.20% Tobacco 21 2.86 0.55 4.34 0.50 2.65% 5.86% 4.74% 2.05% 25.88% 6.71% Food wholesale, retail, and services: 18.76 1.83 7.83 1.12 0.67% 0.94% 4.29% 1.81% 30.45% 5.43% Food wholesalers 514 14.47 1.17 15.57 1.90 1.05% 1.60% 3.44% 1.55% 25.25% 5.88% Food store retail 54 10.02 0.74 6.25 1.25 0.52% 0.76% 5.83% 2.60% 33.48% 5.39% Food service 581 31.78 3.58 1.67 0.22 0.44% 0.47% 3.59% 1.29% 32.61% 5.02% Total food supply chain 12.69 1.61 6.02 0.94 1.11% 1.67% 5.28% 2.49% 27.32% 5.70%

PAGE 46

46 Figure 31. Firm number in FPB industry 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 Firm Number 116 124 122 115 115 116 119 138 147 156 166 163 155 145 131 115 111 110 105 102 97 102 95 0 20 40 60 80 100 120 140 160 180 Firm number in the FPB industry

PAGE 47

47 Figure 32. Firm number in the FWRS industry 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 Firm number 157 156 142 128 126 132 148 167 175 187 203 193 168 154 143 122 118 112 104 102 100 89 78 0 50 100 150 200 250 Firm number in the FWRS industry

PAGE 48

48 Figure 33 EBTO CE fluctuation by weighted average in the FWRS industry 10% 5% 0% 5% 10% 15% 20% 25% 30% 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 Food retailers(54) food wholesalers(514) Food services(581)

PAGE 49

49 Figure 34 EBTOCE fluctuation by weight ed average in the FPB industry 0.15 0.10 0.05 0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 SIC 201 MEAT SIC 202 Diary SIC 203 C.F, and PF and V SIC 204 Grain mill SIC 205 Bakery

PAGE 50

50 Figure 34 Continued 0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.45 0.50 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 SIC 206 Sugar and confectionery SIC 207 Fats and oils SIC 208 Beverage SIC 209 Miscellaneous Food SIC 210 Tobacco

PAGE 51

51 CHAPTER 4 EMPIRICAL RESULTS Before performing the regression, the regression of residuals for the fixed effect panel model is tested. Without the obvious heteroskedastic error and autocorrelation, the model exhibits an overall linear good fit. Regression results for testing the model After checking the residuals of the model, hypothesis 1 proposed in the Chapter 1 is tested. Test hypothesis 1the component financial rat ios of the DuPont expansion have the same effects on the profitability, in terms of EBTOCE, among all sub sectors that comprise of the food supply chain. Considering the different management strategies will be adapted to different industries, r egression t ests are performed separately for the FPB and the FWRS which comprise ten and three sub sectors respectively As stated in the Chapter 2, each DuPont component financial ratio is measuring one proxy variable of sub sectors performance. The twoway fixed effects panel regression is based on the weighted average mean using net sales as the weight. The empirical results are reported in Ta bles 4 1 and 42 for the two main industries Since this research is carried out in two stage s, the crosssection and time effects results will be show n later as part of the testing of hypothes e s 2 and 3. The F test for the twoway fixed effects model focuses only on the presence of the fixed effects, not on the significance of the explanatory variables. Thus, the OLS regression was run to obtain the F statistics for the ten component financial ratios

PAGE 52

52 model. It aims to evaluate the null hypothesis that all regression coefficients are equal to zero versus the alternative that at least one i s not. The results for the food processing and beverage (FPB) industry are shown in Table 4 1 The two way fixed effects regression model for the FPB has an R squared of 0.7182. F statistics for the independent variables in the model is 19.23, with a pvalue of 0.000. This indicates that 71.82% of a change in EBTOCE can be explained by the changes of independent regressors. In other words, the ten sub components financial ratios in this model explain 71.81% of EBTOCEs fluctuation in their linear relationship. The F test from the OLS regression shows the usefulness of this model. The proposed relationship between the EBTOCE and the set of financial ratios is stati stically reliable. The F test shows overall significance of this model. However, not all the ratios are statistically significant. GM/S, IN/S, SG&A/S and R&D/S, proxies of profit margin rate, are all statistically significant at 5% level. In the FPB industry, EBTOCE is very sensitive to the change of profit margin, which is reflected by a coefficient of 0.798. This means, one percent increase in the gross margin rate will increase the EBTOCE by 0.798 percent. Similarly, one percent increase in IN/S, SG&A/ S and R&D/S will decrease the firms EBTOCE by 1.488, 0.581 and 2.329 percent respectively. It is worth mentioning that the research and development expense in agribusiness has a negative impact in its profitability. R&D expenses in the food processing and beverage sector have not been effectively increasing contemporaneous profitability. That might be attributed to the lagged effect of research and development expenditures. The second component tested is the statistical significance of the asset turnover. Under this, there are three ratios ARTR, INTR and PP&ETR. Out of these three, only

PAGE 53

53 PP&ETR (net sales/ average PP&E) is statistically significant at the 5% level. It means that one unit of higher PP&E turnover (annual rate in times) increases EBTOCE by 0.01 percent. Keeping net sales constant, the lower investment in PP&E leads to lower capital employed, which is total assets minus current liability. Thus the higher PP&E turnover rate, the higher is the EBTOCE. The final component of the DuPont expansion is Leverage. Under this, three interest bearing debts are tested: STD/A, CD/A, and LTD/A. The long term debt to assets ratio (LTD/A) is statistically significant, while the other two are not. LTD/A has a coefficient of 0.142 and P value of 0.011 which s hows an inverse relationship with EBTOCE. One percent increase in long term debt results in a decrease in EBTOCE by 0.142 percent. That inverse relationship indicates that the debt to equity capital mix in the current FPB industry should be reduced to improve return on capital employed. The results for the f ood w holesale, r etail and service (FWRS) industry are provided in Table 4 2 The regression model with respects to the FWRS has an R squared of 0.7955 and an F statistics of 9.77 with a probability val ue of 0.00. It means that the model measures 79.55 % of EBTOCEs variability. In contrast with the FPB, only interest expense over sales (IN/S) and long term debt to assets ratio (LTD/A) are statistically significant at 5% level, with a coefficient of 4. 46% and 0.56%. Debt load is significantly affecting the profitability of FWRS inversely. One percent increase in LTD/A ratio will result in a decrease in EBTOCE by 0.56% and one percent increase in interest expense per dollar of sales will decrease EBTOC E by 4.46%. Besides above ratios, GM /S and INTR are significant at 15% level, and STD/A is significant at 10% level It is surprising to see the statistical insignificance of other

PAGE 54

54 financial ratios. It might be attributed to the relative small business scale in FWRS. Table 31 sh ows that there are, on average, 125 firms in ten sub sectors of FPB, while there are139 firms in the three sub sectors of FWRS. Especially, the food services sub sector (SIC 581) has, on average, 89 firms per year. Many of those restaurants and food providers are small scale enterprises, which might be managed differently than large firms. The high variability in production planning and action strategies for these small businesses makes it difficult to predict their profitability using a specific model. T he change in EBTOCE is not majorly reflected by their f inancial ratios. Regression results for testing the crosssection and time effects The two way fixed effect panel mod el is constructed to test hypothes e s 2 and 3 at the same time. Test hypothesis 2: EBTOCE does not have cross section effects within the U.S. food supply chain. All sub sectors profitability levels have no signifi cant difference statistically. Test hypoth esis 3: Sub sector s EBTOCE does not vary across time. There are no uniform time effects to those sub sectors. The null hypothesis is that parameters of sub sectors and time dummies are zero. It is a twofold test. The first test is the cross sectional eff ect test. It captures the sub sector effects which are constant over time. In this study, this refers to the within differences of the component sub sectors in the FPB and FWRS The second test aims to capture the differences over 23 years that is common to all component sub sectors in the U.S. food supply chain and to explore whether the time effect pattern follows the recent three economic recessions. As stated in the Chapter 2, to avoid the perfect

PAGE 55

55 multicollinearity ( the dummy variable trap), the model drops one cross section and one time series dummy variables in a twoway fixed effect model. Table 4.3 describes the basic output of the twoway fixed effects model for the food processing and beverage industry Panel A shows that there are ten cross sections and twenty three time observations. Ten cross se ctions refer to the 3 digit SIC code that starts with 20 and the tobacco sub sector with SIC code 21. The 23 time observations cover the periods from 1986 to 2008. The F test for the fixed effects shown in Table 4 3 Panel B, tests the null hypothesis that there are no fixed effects. The P value for the F statistic is almost zero. So we reject the null hypothesis which indicates that we cannot use OLS to estimate the parameters. There are sub sector e ffects, time effects, or both. The test is highly significant. To explore the detail of fixed effects, two separate oneway fixed effect regressions are developed regarding the sub sector difference and time impact respectively. Table 43 Panel C provides the results of two separate F tests, their small P values are indicating the existence of both sub sector effects and time effects. Table 44 presents the twoway fixed effects regression results with respect to FPB. There are ten cross section and 23 t im e series in that model. Variables d1, d2 to d10 are assigned to ten sub sector dummies; year dummies. SAS sorts the dummy variables by an ascending order and drop the last crosssection dummy and time dummy automatic ally. The 230 regression equations (23x10) can be drawn on the combinations of ten sub sectors and 23 years. The twoway fixed effect panel regression model assumes that independent variables have constant slopes, only intercepts vary according to cross section and time. The

PAGE 56

56 parameter estimate of sub sector is the intercept ( 0.038), a reference point. The other dummy parameter coefficients are computed using this reference point as follows: intercept+ sub sector dummy coefficient + year dummy coefficient+ other dummy coefficients*0. The actual intercept of meat sub is computed as intercept+d1 coefficient + 22 coefficient +other dummy estimates*0. That is 0.043= 0.038+ 0. 06+ 0. 021+0. The coefficient 0.043 means, holding other explanatory variables constant, the EBTOCE of meat sub sector in 2007 is 4.3% higher than tobacco sub sector in 2008. The coefficients of other sub sector s years can be estimated similarly. L ooking at the parameters in Table 4 4, most of the cross section effects are highly significant (with the exception of sub sectors 20 2 205, 20 7 and 208). Tobacco sub sector ( SIC 21 ) as a controlling group is sorted and dropped by SAS as a default. The sm all P value of F statistics means that the other sub sectors are significantly different from the tobacco sub sector A significant influence of sub sector specific factors is present. Even thought in 12 out of 23 time periods the time effect is statistically significan t, the t ime s eries e ffects have no uniform pattern in FPB. It is shown that t here are some significant fluctuation s occurred in the past 23 years, but not all sub sectors experienced the significant decrease in return in a same cer tain year. Holding other independent variables constant, the profitability of a sub sector in a specific recession period does not exhibit the significant decrease in return comparing with other time periods. From year 2000, however, the year dummy variables lose their significance. The parameters values decrease in size. The interesting fact is, the visible time effects pattern do not

PAGE 57

57 match with the three economic recessions during the observation range. Corresponding to the fluctuation analy sis of data in the Chapter 3, the change of EBTOCE attributes to the management strategy of agribusiness rather than economic downturns Generally, the overall demand shrinks along with the recession. However, it seems to have no statistical ly significant impact to the FPB industrys profitability in terms of EBTOCE The two way fixed effects model output is provided in Table 4 5 for FWRS industry Panel A shows that there are three cross sections and 23 time observations. Three cross sections are sub sector SIC 54, SIC 514 and SIC 581. Identical with previous definition, 23 time observations cover the periods from 1986 to 2008. The F test for fixed effects is shown in the Table 4 5 Panel B provides evidence that th ere are no fixed effects. With t he F statistic of 1.27 and associated P value of 0.259, the null hypothesis is not reject ed. There are no sub sector effects, or time effects. To explore the details of fixed effects, two separate oneway fixed effect regression models are developed regard ing the sub sector difference and time impact respectively. Table 45 Panel C reports two separate F test. Both the p value for the F Statistic is above 10% significant level, which indicates an absence of any sub sector effect and time effects. The sub se ctor dummies and year dummies are not statistically significant in this industry In essence, there is no statistical difference in EBTOCE among sub sectors in FWRS over the past 23 years. The statistic output of FWRS in the Table 4 6 corroborates this conclusion. Twenty of the P values of coefficient estimates are greater than 5%. That result might be attributed to the fact that there are too many small scale agribusinesses in FPB groups, like groceries and restaurants. They are more variable and complex i n profitability than large companies. There might be no common operating

PAGE 58

58 practice or a specific time period that are influencing those small businesses persistently.

PAGE 59

59 Table 41 Two way fixed effects panel regression in FPB industry Model Description and Estimation Total sample 230 Degree of freedom 186 Fit Statistics (F statistics: 19.23, p value: 0.000) R Square 0.7182 MSE 0.0013 Parameter Estimates Estimate Standard Error t Value Pr > |t| Intercept 0.038 0.056 0.690 0.494 GM/S 0.798 0.113 7.040 <.0001 IN/S 1.488 0.508 2.930 0.004 SG&A/S 0.581 0.141 4.110 <.0001 R&D/S 2.329 0.990 2.350 0.020 ARTR 0.002 0.002 0.810 0.419 INTR 0.001 0.002 0.390 0.698 PP&ETR 0.010 0.004 2.760 0.006 STD/A 0.013 0.125 0.100 0.921 CD/A 0.055 0.099 0.560 0.579 LTD/A 0.142 0.055 2.570 0.011 Table 4 1 presents the statistical output of the regression for the independent variables in FPB. Since the hypothesis 1, 2, and 3 are tested separately, the twoway fixed effects F test and results are moved to Table 4 3. The regression takes the form that LTD/A (it)+ ui+ vt+ Independent variables are defined as following: GM/S is gross margin rate; SG&A/S is selling and general and administrative Expense rate; IN/S is Interest Expenses rate; R&D/S is research and development expense rat e; ARTR is accounts receivables turnover rate; INTR is inventory turnover rate; PP&ETR is property, plant and equipment turnover rate; STD/A, CD/A and LTD/A represent short term debt, current portion of long term debt and longterm debt to assets ratio res pectively.

PAGE 60

60 Table 4 2 Two way fixed effects panel regression in FWRS industry Model Description and Estimation Total sample 69 Degree of freedom 33 Fit Statistics (F statistics 9.77; p value: 0.00) R Square 0.7955 MSE 0.0011 Parameter Estimates Estimate Standard Error t Value Pr > |t| Intercept 0.141 0.211 0.670 0.510 GM/S 1.337 0.846 1.580 0.124 IN/S 4.460 1.378 3.240 0.003 SG&A/S 0.600 1.119 0.540 0.596 R&D/S ARTR 0.000 0.001 0.350 0.728 INTR 0.008 0.005 1.500 0.142 PP&ETR 0.001 0.007 0.200 0.843 STD/A 1.824 0.970 1.880 0.069 CD/A 0.038 0.370 0.100 0.919 LTD/A 0.558 0.138 4.050 0.000 Table 4 2 presents the statistical output of the regression for the independent variables in FWRS INDUSTRY. Since the hypothesis 1, 2, and 3 are tested separately, the twoway fixed effects F test and results are moved to Table 4 3. The regression takes the form (it)+ LTD/A (it)+ ui+ vt SG&A/S is selling and general and administrative Expense rate; IN/S is Interest Expenses rate; R&D/S is research and de velopment expense rate; ARTR is accounts receivables turnover rate; INTR is inventory turnover rate; PP&ETR is property, plant and equipment turnover rate; STD/A, CD/A and LTD/A represent short term debt, current portion of long term debt and longterm deb t to assets ratio respectively. R&D/S coefficient estimate and standard error are 0 because there is no R&D expense in FWRS industry.

PAGE 61

61 Table 4 3 Hypothesis test for fixed cross section effects and time effects in the FPB Panel A: Model Description Estimation Method Fix Two Number of Cross Sections 10 Time Series Length 23 Table 4 3 provides the regression results of cross LTD/A (it)+ ui+ vt

PAGE 62

62 Table 4 3 Continued Panel B : Fixed effect F test Model Fixed eff ect Sample Size F Test Prob value Result Two way Both sectors and time: I =201, 202, 203, 21 T =1986, 1987, 1988..,2008 230 F(31,186)=5.15 <.0001 Null hypothesis for both sector and time effect are not rejected Table 4 3 provides the regression results of cross LTD/A (it)+ ui+ vt

PAGE 63

63 Table 4 3 Continued Panel C: separate F test for individual effect Model Fixed effect Sample Size F Test Prob value Result One way Single sectors effect I =201, 202, 203, 21 230 F(9,208)= 9.21 <.0001 Null hypothesis for sector effect is not rejected One way Single time effect T =1986, 1987, 1988..,2008 230 F(22, 195)=2.82 <.0001 Null hypothesis for time effect is not rejected Table43 provides the regression results of crossLTD/A (it)+ ui+ vt

PAGE 64

64 Table 4 4 Parameter e stimates in the fixed effects model for the FPB Parameter Estimates Variable Variable Estimate Standard Error t Value Pr > |t| Label Sub Sector 201 d1 0.060 0.030 2.010 0.046 Cross Sectional Effect 1 Sub Sector 202 d2 0.046 0.027 1.680 0.094 Cross Sectional Effect 2 Sub Sector 203 d3 0.070 0.017 4.040 <.0001 Cross Sectional Effect 3 Sub Sector 204 d4 0.092 0.018 5.130 <.0001 Cross Sectional Effect 4 Sub Sector 205 d5 0.025 0.022 1.130 0.261 Cross Sectional Effect 5 Sub Sector 206 d6 0.057 0.019 2.960 0.003 Cross Sectional Effect 6 Sub Sector 207 d7 0.056 0.031 1.820 0.070 Cross Sectional Effect 7 Sub Sector 208 d8 0.012 0.019 0.620 0.539 Cross Sectional Effect 8 Sub Sector 209 d9 0.046 0.020 2.260 0.025 Cross Sectional Effect 9 Y ear 1986 0.061 0.019 3.200 0.002 Time Series Effect 1 Y ear 1987 0.098 0.019 5.170 <.0001 Time Series Effect 2 Y ear 1988 0.063 0.019 3.380 0.001 Time Series Effect 3 Y ear 1989 0.059 0.019 3.040 0.003 Time Series Effect 4 Y ear 1990 0.060 0.019 3.170 0.002 Time Series Effect 5

PAGE 65

65 Table 4 4 C ontinued Parameter Estimates Variable Variable Estimate Standard Error t Value Pr > |t| Label Year 1991 0.053 0.020 2.690 0.008 Time Series Effect 6 Year 1992 0.044 0.019 2.270 0.024 Time Series Effect 7 Year 1993 0.035 0.019 1.810 0.072 Time Series Effect 8 Year 1994 0.037 0.019 2.010 0.046 Time Series Effect 9 Year 1995 0.056 0.019 2.970 0.003 Time Series Effect 10 Year 1996 0.040 0.019 2.150 0.033 Time Series Effect 11 Year 1997 0.024 0.018 1.340 0.183 Time Series Effect 12 Year 1998 0.014 0.018 0.760 0.448 Time Series Effect 13 Year 1999 0.045 0.019 2.390 0.018 Time Series Effect 14 Year 2000 0.025 0.019 1.320 0.190 Time Series Effect 15 Year 2001 0.020 0.018 1.080 0.282 Time Series Effect 16 Year 2002 0.017 0.018 0.960 0.339 Time Series Effect 17 Year 2003 0.001 0.018 0.030 0.976 Time Series Effect 18 Year 2004 0.008 0.018 0.440 0.659 Time Series Effect 19 Year 2005 0.018 0.018 1.030 0.302 Time Series Effect 20 Year 2006 0.003 0.018 0.180 0.860 Time Series Effect 21 Year 2007 0.021 0.017 1.200 0.233 Time Series Effect 22 Sector 21 in 2008 0.038 0.056 0.690 0.494 Intercept Ten explanatory variables parameter estimates are shown in table4 1 SIC groups classifications follows: Meat (201); diary (202); canned, frozen, and preserved fruits and vegetables (203); grain mill (204); bakery (SIC 205); sugar and confectionery (206); fats and oils (207); beverages (208);miscellaneous food preparations and kindred (209); tobacco (21); food service (5810 and 5812); retailers (5400 and 5411); and wholesalers (5140, 5141, and 5180).

PAGE 66

66 Table 4 5 Hypothesis test for the fixed crosssection effects and time effects in the FWRS industry Panel A: Model Description Estimation Method Fix Two Number of Cross Sections 3 Time Series Length 23

PAGE 67

67 Table 4 5 Continued Panel B : Fixed effect F test Model Fixed effect Sample Size F Test Prob value Result Two way Both sectors and time: I =54,514,581 T =1986, 1987, 1988..,2008 69 F(24,33)= 1.27 0.2588 Null hypothesis of both sector and time effect are not rejected

PAGE 68

68 Table 4 5 Continued Panel C: separate F test for individual effect Model Fixed effect Sample Size F Test Prob value Result One way Single sectors effect I = 54,514,581 69 F(2,55)= 1.57 0.2167 Null hypothesis for sector effect is not rejected One way Single time effect T =1986, 1987, 1988..,2008 69 F(22, 35)=1.21 0.3079 Null hypothesis for time effect is not rejected Table45 provides the regression results of cross LTD/A (it)+ u i+ vt

PAGE 69

69 Table 4 6 Parameter estimates in the fixed effects model for FWR S industry Parameter Estimates Variable Variable Estimate Standard Error t Value Pr > |t| Label Sub Sector 54 d1 0.151 0.176 0.860 0.398 Cross Sectional Effect 1 Sub Sector 514 d2 0.175 0.106 1.650 0.109 Cross Sectional Effect 2 Year 1986 0.092 0.067 1.360 0.182 Time Series Effect 1 Year 1987 0.116 0.067 1.740 0.090 Time Series Effect 2 Year 1988 0.129 0.050 2.590 0.014 Time Series Effect 3 Year 1989 0.153 0.052 2.970 0.006 Time Series Effect 4 Year 1990 0.179 0.056 3.200 0.003 Time Series Effect 5 Year 1991 0.123 0.047 2.650 0.012 Time Series Effect 6 Year 1992 0.083 0.039 2.130 0.041 Time Series Effect 7 Year 1993 0.061 0.039 1.550 0.132 Time Series Effect 8 Year 1994 0.052 0.043 1.220 0.232 Time Series Effect 9 Year 1995 0.030 0.036 0.840 0.407 Time Series Effect 10 Year 1996 0.024 0.036 0.660 0.511 Time Series Effect 11 Year 1997 0.009 0.032 0.270 0.789 Time Series Effect 12 Year 1998 0.015 0.034 0.430 0.670 Time Series Effect 13 Year 1999 0.009 0.031 0.310 0.762 Time Series Effect 14 Year 2000 0.008 0.031 0.270 0.790 Time Series Effect 15 Year 2001 0.020 0.028 0.730 0.468 Time Series Effect 16 Year 2002 0.023 0.028 0.830 0.410 Time Series Effect 17 Year 2003 0.001 0.029 0.030 0.976 Time Series Effect 18 Year 2004 0.018 0.029 0.630 0.533 Time Series Effect 19 Year 2005 0.010 0.030 0.320 0.749 Time Series Effect 20 Year 2006 0.012 0.030 0.410 0.683 Time Series Effect 21 Year 2007 0.011 0.027 0.420 0.680 Time Series Effect 22 Sector 581 in year 2008 d3 0.141 0.211 0.670 0.510 Intercept T en explanatory variables parameter estimates are shown in table4 2 SIC groups classifications follow: meat (201); diary (202); canned, frozen, and preserved fruits and vegetables (203); grain mill (204); bakery (SIC 205); sugar and confectionery (206); fats and oils (207); beverages (208);miscellaneous food preparations and kindred (209); tobacco (21); food service (5810 and 5812); retailers (5400 and 5411); and wholesalers (5140, 5141, and 5180).

PAGE 70

70 CHAPTER 5 CONCLUSION T his research aims to examine the U.S. agribusinesss profitability based on the decomposition of the DuPont system with consideration of the recent three U.S. economic recessions. The overall approach is to derive the fin ancial variables that stem from the DuPont formula as independent variables, t o measure the dependent variable which is the return on capital employed for agribusiness es The effects of different financial aspects to EBTOCE are tested by this regression. A dditionally, the cross sectional effects and time effects of the U.S. food supply chain over past 23 years are tested Through the analysis of profitability fluctuation, this study provides an insight into the agribusiness financial performance. This research is a financial statement analysis based on the DuPont system with respect to U.S. food supply system. Three hypotheses are tested with respect to two major food industries and subordinate 13 sub sector s. In terms of the first hypothesis test, t he empi rical result indicates th at agribusiness profitability among all sub sectors is negatively affected by long term debt. The higher the long term debt to asset mix, the lower return on capital employed. That conclusion can be applied to both FPB and FWRS. Furthermore, the profit margin (including four major components: GM/S, SG&A/S IN/S, R&D/S) have significant impact on the profitability of all the sub sectors in the FPB industry However, except for the IN/S, other profit margin components which are prox ies of the firms operating efficiency exhibit statistical insignificance in the FWRS industry A ssets turnover ratios have three major components: ARTR INTR and PP&ETR. Except the PP&ETR in FPB, c oefficient estimations of other assets turnover rati os provide evidence that assets usage efficiency

PAGE 71

71 is not statistically affecting the EBTOCE for almost all sub se ctors in the food supply chain. In other words, the profitability change, in terms of EBTOCE, cannot be explained by the variations of assets turnover rate. In terms of the second and third hypotheses, the two different industries have different test results. S ignificant crosssection effect s are present in the FPB industry It indicates that the profitability is significantly different across sub sectors in that industry. However, FWRS s profitability is less relevant to its sub sector s, because the crosssection effect is not report ed from the panel data regression statistical test. It is shown that the economic downturns do have an impact on the number of market participants. The reduced number of market participants is highly associated with economic recessions. Especially small agribusinesses which drop out the market file for bankruptcy or merge. However, in th is model, the time effect h as been tested on surviving firms. Even though there are significant time effect s exhibits in the FPB the fluctuation of EBTOCE is not associat ed with recession periods, which indicates that there are other factors or reasons that are playing a more important role in the fluctuation of firms profitability EBTOCE havent been affected materially by the well recognized recession periods. A decrease in profitability might stem from various factors or reasons. By analyzing the individual firms performanc e closely, it is found that decreas es in return are mainly attribut ed to ineffective or inefficient management, such as fail to control manufacturing cost, bad acquisition, fail ure to hedge the foreign exchange risk, stronger competitors etc. Ultimately, this study concludes that the profitability of the food supply chain is mainly affected by management operating

PAGE 72

72 strategy. Economic recessions do not have significant impact on profitability of the U.S. food supply chain statistically. The major research limitation of this study was the failure to collect the market variables, such as stock prices, price earnings rates (PE) and earnings per share (EPS) These a re indicator s of a company's profitability as perceived by investors. That is also the drawback of DuPont formula, which is the main focus on the internal measurement of firms performance.

PAGE 73

73 LIST OF REFERENCES Banker, R., Chang, H. S., and Majumdar, S. K. 1993. Analyzing the underlying dimensions of firm profitability. Managerial and Decision Economics 14 : 25 36. Bourlakis, M. A., and Weightman, P. W. 2004. Food Supply Chain Management Wiley Blackwell. Fairfield, P., J. Whisenant, and T. Yohn. 2003. The differential persistence of accruals and cash flows for future operating income versus future profitability. The Review of Accounting Studies 8: 221243. Forster, D. L. 1996. Capital structure, busines s risk, and investor returns for agribusinesses. Agribusiness 12 : 429 442. Freeman, R. J., and Penman, S. 1982. Book rateof return and prediction of earnings changes: An empirical investigation. Journal of Accounting Research 20 : 639653. Jensen, M. C. 1989. The eclipse of the public corporation. Harvard Business Review 5, 61 74. Magdoff, H., Foster, J. B., and Buttel, F. H. 2000. Hungry for Profit: The Agribusiness Threat to Farmers, Food, and the Environment. Monthly Review Press. Ne ibergs, J. S. 1998. Macroeconomic Conditions and Agribusiness Profitability: An Analysis Using Pooled Data. International Food and Agribusiness Management Review 1 : 91 105. Nissim, D. and S.H. Penman. 2001. Ratio analysis and equity valuation: From research to practice. Review of Accounting Studies 6: 109154. O pler T. C., and S heridan, T. 1994. Financial Distress and Corporate Performance. The Journal of Finance 49: 10151040. Selling, T., and Stickney, C. 1989. The effects of business environments and strategy on a firms rate of return on assets. Financial Analysts Journal 45 : 43 52. Trejo Pech, C. O., Weldon, R. N., and House, L. A. 2008. Earnings, accruals, cash flows, and EBITDA for agribusiness firms. Agricultural Finance Review 68 : 301 319. Yaffee, R. 2003. A primer for panel data analyses. http://www.nyu.edu/its/pubs/connect/fall03/yaffee primer.html last accessed July 2010.

PAGE 74

74 BIOGRAPHICAL SKETCH Youshan Zhao was born in Beijing and obtained her bachelors degree in a ccounting from North China University of Technology, Beijing, China in 2003. Miss. Zhao is China Certified Public Accountant and worked for PRICEWATERHOUSECOOPERS Beijing, China for three and a half years. The requirements for the degree of Master of Science in food and resource economics with a minor in statistics were completed in August 2010 at t he University of Florida.