Environmental and Resource Economics 9: 323-340, 1997.
(~) 1997 Kluwer Academic Publishers. Printed in the Netherlands.
323
Do Farmers Overuse Nitrogen Fertilizer to the
Detriment of the Environment?
SATYA N. YADAV l , W I L L I S P E T E R S O N 2 and K. W I L L I A M E A S T E R 2
i Department of Agricultural Economics & Rural Sociology, 221, Agricultural Building, University
of Arkansas, Fayetteville, AR 72701, USA; 2 Department of Applied Economics, University of
Minnesota, St. Paul, MN 55108, USA
Accepted 9 July 1996
Abstract. Increasing use of nitrogen fertilizer in U.S. agriculture has led to nitrate contamination
of water resources. The main objective of the study is to determine if the current use of nitrogen
exceeds the profit-maximizing level, since reducing such discrepancy, if any, could raise farmers'
profitability and enhance water quality making it a win-win situation. Specific objectives of the study,
however, are two-fold. First, develop an appropriate methodology for estimation of an agronomic
production function utilizing panel data with several treatments from experimental plots. Second,
using experimental data from 1987 through 1990 for three farm sites in southeastern Minnesota,
empirically estimate the production function and profit maximizing level of nitrogen application.
Our results show that both the current recommended rate, 150 lb/acre, and farmers' use, 176
lb/acre, of nitrogen exceed the profit maximizing level of nitrogen in the region. It is shown that the
recommended rate needs to be revised and made more site or area specific rather than a general figure
for the entire region. The study shows considerable residual nitrogen build-up in the soil profile,
implying that farmers have been applying more nitrogen than could be utilized by corn plants. The
later finding highlights the importance of soil testing for plant nutrients when making decisions on
fertilizer application.
Key words: ground water contamination, nitrate contamination, nitrogen overuse, profit maximization
1. I n t r o d u c t i o n
The d e v e l o p m e n t o f high-yielding crop varieties and a reduction in chemical and
fertilizer prices h a v e greatly increased the use o f these inputs in U.S. agriculture.
C o n s u m p t i o n o f primary nutrients (nitrogen, phosphorus, and potash) has increased
b y m o r e than threefold, and that o f nitrogen m o r e than fourfold between 1960 and
1990 (Vroomen 1989). Increased fertilizer use, however, has not c o m e without
costs or externalities to society (Johnson et al. 1989)J Besides soil erosion and
surface water pollution, the p r o b l e m o f ground water contamination b y nitratenitrogen is b e c o m i n g a serious and ubiquitous environmental concern (Pye 1983).
Indeed, such contamination m a y result from several sources, including industrial
wastes, municipal landfills, mining activities, septic systems, and agricultural activities. However, evidence suggests that agriculture, through the usage o f inorganic
fertilizers, is a m a j o r contributor (Hallberg et al. 1985).
324
SATYA N. YADAV ET AL.
Ground water contamination by nitrate-nitrogen has been confirmed in 31 states
and is suspected in 6 others (Lee and Nielsen 1987). In Minnesota, the percentage
of wells exceeding EPA's standard varies from 1% to 44% across data sets, for the
period between 1978 and 1990. 2 Nevertheless, EPA analysis found that 7.3% of
wells exceeded 10 mg/1 of NO3-N. This figure can be regarded as a good approximation of the incidence of ground water contamination problems in Minnesota
(Wall 1991). Concerns over the pollution of ground water and/or surface water arise
because of their expected adverse effects on human and livestock health as well as
on the environment, including methemoglobinemia or 'blue baby' syndrome.
Throughout the state of Minnesota, 75% of the population depends on ground
water sources for their drinking water supply. Moreover, ground water is used
by 93% the state's municipal water supply systems. In addition to these vital
uses, ground water is essential in industrial usages, including food and beverage
processing, and livestock production (MPCA 1989). Given the dependence on
ground water, nitrate contamination of ground water is causing growing anxiety at
both the state and federal levels.
The main objective of the study is to determine if farmers' use of nitrogen in
southeastern Minnesota exceeds that of the profit-maximization level. If it can be
shown that the current nitrogen application rate is higher than the optimal level
in a (non-environmental) profit-maximizing context, the reduction in fertilizer use
would increase farm profitability and enhance water quality. This is a win-win
situation. Specific objectives of the study, however, are twofold. First, we develop
an appropriate methodology for estimation of an agronomic production function
utilizing panel data with several treatments from experimental plots. Second, using
experimental data from 1987 through 1990 for three farm sites in southeastern
Minnesota, we empirically estimate the production function and profit-maximizing
level of nitrogen application. This is then compared with the current level of
nitrogen usage in the region.
2. Experimental Design and Data Collection
The data for this study are from experimental sites on three farms: Site 1 (in
Olmsted county), Site 2 (in Winona county), and Site 3 (in Goodhue county).
All lie in southeastern Minnesota where nitrate contamination of ground water
is a growing concern. Well samples taken in the mid-1980s by the Minnesota
Department of Health and the Minnesota Department of Agriculture indicated that
39% of the rural private wells in this area exceeded the U.S. public health drinking
water standard of 10 mg NO3-N/1 (Randall et al. 1990).
There could be a couple of reasons for such an increased level of contamination
in this part of the state. First, the intensive agricultural production system in the
region uses a considerable amount of chemical fertilizers and manures. Second,
soils in southeastern Minnesota are rather shallow over a fractured limestone and
sandstone bedrock, called Karst. Mature Karst structure is characterized by sink-
DO FARMERSOVERUSENITROGENFERTILIZER?
325
Table I. Description of nitrogen and tillage managements at three experimental sites
Treatments
type
Applications of nitrogen (lb/acre)
Site 1 (Olmsted Co.)
Site 2 (Winona Co.) Site 3 (Goodhue Co.)
No. 1
No. 2
No. 3
No. 4
No. 5
No. 6
No. 7
No. 8
No. 9
No. 10
No. I 1
No. 12
No. 13
No. 14
No. 15
No. 16
0 (CP)
75 (CP, spring)
150 (CP, spring)
255 (CP, spring)
t50 (CP, fall)
150 (CP, fall, NH)
150 (CP, spring, SD)
150 (NT, spring)
175 (CP, spring, manure) a
310 (CP, spring, manure) a
0 (CP)
50 (CP, PP)
100 (CP, PP)
150 (CP, PP)
200 (CP, PP)
0 (NT)
100 (NT, PP)
150 (NT, PP)
200 (NT, PP)
100 (CP, PP + SD)b
150 (CE PP + SD)c
150 (CE SD 4-1F)
0 (CP)
50 (CP, PP)
100 (CP, PP)
150 (CP, PP)
200 (CP, PP)
0 (NT)
100 (NT, PP)
150 (NT, PP)
200 (NT, PP)
100 (CP, PP + SD)b
150 (CE PP + SD)c
150 (CE PP + SD)a
100 (CP, SD 4-1F)
150 (CE SD 4-1F)
150 (CE PP, NH)
150 (CE SD41E NH)
a Liquid hog manure applied only in 1987 and 1988. No manure or fertilizer was applied
in 1989 or 1990.
b p p + SD: 50 lb nitrogen before planting and 50 lb after planting.
c p p + SD: 50 lb nitrogen before planting and 100 lb after planting.
d p p + SD: 100 lb nitrogen before planting and 50 lb after planting.
CP: Tillage with chisel plow; NT: no tillage; spring/fall: nitrogen applied in spring/fall;
SD: nitrogen applied in split dose; SD 4-1F: nitrogen applied at 4-leaf stage; PP: nitrogen
applied before planting; NH: use of nitrogen inhibitor.
f Unless stated, source of nitrogen is anhydrous ammonia, split dose implied half before
and half after planting at knee high stage of plant growth.
holes, enlarged joints, n u m e r o u s springs, disappearing streams, c a v e systems, and
dry valleys. Often, such t o p o g r a p h y has p o o r surface drainage, thereby allowing
a quick m o v e m e n t o f water into the sub-soil through sinkholes and disappearing
streams. Thus, ground water in Karst terrain is closely related to surface reservoirs
(Jannik et al. 1991; Growl 1986).
2.1. EXPERIMENTAL DESIGN
T h e data were collected by the D e p a r t m e n t o f Soil Science, University o f Minnesota. T h e experimental procedures for collecting data f r o m the three sites were
initiated in 1986. However, actual data collection was started in 1987 and lasted
through 1990. Altogether, there were 38 treatments, 10 at site 1, 12 at site 2, and
16 at site 3, each planted with continuous corn (see Table I for details). Continuous corn was selected as the cropping pattern because it is a c o m m o n practice in
326
SATYAN. YADAVET AL.
Table II. Distribution of residual nitrogen (lb/acre) in the soil profile across the
three sites
Depth (feet)
Ammonium-nitrogen(NH4-N)
Site 1 Site 2 Site 3
Nitrate-nitrogen (NO3-N)
Site 1 Site 2 Site 3
0-1
1-2
2-3
3--4
4--5
5--6
6-7
7-8
16.7
5.7
5.3
3.3
2.8
3.3
3.8
2.2
18.7
8.2
3.9
2.7
2.4
2.4
3.6
4.1
18.4
8.7
10.8
9.9
8.7
7.6
7.7
6.6
21.5
15.4
10.5
8.7
4.6
4.5
5.7
4.4
76.4
61.9
66.4
52.7
32.9
24.8
18.6
17.4
23.1
17.3
14.8
14.5
23.2
22.3
23.1
18.7
Note: Figures are based on 1987 sampling.
the region (MDA 1991). Moreover, corn utilizes a large share of the nitrogenous
fertilizers applied in the region (USDA 1991).
A randomized complete-block design with four replications was established at
each site, for every treatment, every year.3 Accordingly, the data set consists of
608 (38 x 4 x 4) observations. Treatments among and within sites varied from
each other only with respect to fertilization and tillage management practices. All
other aspects of cultivation practices were controlled, except weather. Fertilizer
and tillage management practices were varied in the experiment in order to see
their impact on ground water pollution. Variations in fertilization included time
of application (fall vs. spring; pre-plant vs. four-leaf stage), method of application
(split vs. single dose), with and without use of nitrogen inhibitor, and source
of nitrogen (anhydrous ammonia vs. manure). The variation in tillage practices
included either deep tillage using chisel plow or no tillage. In addition, there were
several treatments at each site with nitrogen rates of 150 lb/acre, the extension
service recommended rate for corn in the region. Treatments involving this rate
along with variations in cultivation practices were included in the experiment to
examine their effects on yields and the residual nitrogen build-up in the soil profile.
2.2. RESIDUAL NITROGEN IN SOIL PROFILE
The data set collected for the three sites includes information on corn yield
(bu/acre), fertilizer use, and nitrate concentration in the soil profile associated
with each replication as well as other factors. Nitrate-nitrogen concentrations in
lb/acre were recorded for each 1 foot incremental depth of soil reaching down
to 8 feet depth. Table II shows layer-wise average residual nitrogen measured in
lb/acre, across the three sites. The data were collected in April, 1987 (i.e. prior
to the start of the experiment) and can be regarded as an indication of nitrogen
build-up in the soil profile of the respective areas. In general, site 2 has the highest
DO FARMERSOVERUSENITROGENFERTILIZER?
327
Table IlL Residual nitrate-nitrogen (lb/acre) at 5 feet depth of soil profile under various
nitrogen application levels
Month and
Surface applications ofnitrogen (lb/acre)
year
0
50
75
100
150
200
225
Site 1
1987 Oct.
1988April
1988 Oct.
1989April
1989 Oct.
1990April
1990 Nov.
62
50
26
63
27
32
11
-
73
62
47
83
43
132
38
-
99
118
151
182
147
200
124
-
213
154
241
208
325
416
192
Site2
1987Oct.
1988April
1988 Oct.
1989April
1989 Oct.
1990April
1990Nov.
152
129
104
199
65
79
31
191
128
119
182
105
81
53
-
270
191
215
250
158
152
129
347
195
289
351
308
363
206
447
311
414
400
641
412
366
Site 3
1987 Oct.
1988 April
1988 Oct.
1989 April
1989 Nov.
1990 May
1990 Nov.
49
47
43
58
17
36
31
63
NA
218
NA
22
NA
46
-
80
NA
120
NA
68
NA
67
106
99
269
196
108
148
142
181
139
382
199
263
237
249
-
D
D
D
NA: not available.
n i t r o g e n b u i l d - u p f o l l o w e d b y the sites 3 a n d 1, r e s p e c t i v e l y . A s e x p e c t e d , nitrog e n a c c u m u l a t i o n d e c r e a s e d w i t h i n c r e a s i n g d e p t h o f the soil profile. T a b l e III
s h o w s y e a r - w i s e n i t r a t e - n i t r o g e n a c c u m u l a t i o n ( n o n - a c c u m u l a t i v e ) at 5 feet d e p t h
o f the soil profile, a s s o c i a t e d w i t h v a r i o u s l e v e l s o f s u r f a c e n i t r o g e n a p p l i c a t i o n s .
A g r o n o m i c studies r e v e a l that n i t r a t e - n i t r o g e n at o r b e l o w 5 feet d e p t h c a n n o t b e
u s e d b y the c o r n p l a n t a n d thus is c o n s i d e r e d to b e a p o t e n t i a l s o u r c e o f g r o u n d
w a t e r p o l l u t i o n . N i t r o g e n is m o s t r e a d i l y a v a i l a b l e to c o m p l a n t s in the 0 to 2 o r 3
feet depth, d e p e n d i n g u p o n m o i s t u r e c o n t e n t o f the soil profile. T h e data r e v e a l a n
i n c r e a s e in nitrate a c c u m u l a t i o n in the soil profile with i n c r e a s i n g l e v e l s o f n i t r o g e n
applications.
328
SATYA N. YADAV ET AL.
3. Agronomic Yield Functions: Previous Studies
Output requires inputs. Inputs can be controllable, such as fertilizers, pesticides,
labor, and machinery, or uncontrollable, including rainfall, soil type, and topography. Assuming that all inputs, except macronutrients, are held constant, a general
form of the corn response function can be represented as:
Y = f(P IX),
(1)
where Y is corn yield in bushels per acre, P is a vector of macronutrients, each
measured in pounds per acre, and X is a vector of all other inputs that are held
constant. It becomes a critical task to adopt the correct functional form which
describes the production relationship.
Notable early works on yield functions include those of Spillman (1923), Heady
(1952), Johnson (1953), Heady and Dillon (1961), Perrin (1976), and Heady and
Hexem (1978). These studies empirically investigated the characteristics imposed,
a priori, by the chosen functional form. De Janvry (1972) presented a model for
optimal fertilizer use that accounted for risk due to weather variability.
Griffin et al. (1987) identified 20 traditional and popular functional forms, and
discussed their intrinsic properties. In order to choose a best form for a given
task, they proposed to choose one over the other based on maintained hypotheses
(homogeneity, homotheticity, elasticity of substitution, and concavity), ease of estimation (availability and properties of data, the resources available for computation,
etc.), availability of data (goodness of fit and general conformity to data), and/or
expected application of the results.
Judging from the recent literature, the most commonly used agronomic response
functions appear to be: quadratic, von Liebig, and the Mitscherlich-Baule (Frank
e t a . 1990). Their algebraic formulations and characteristics are discussed here
briefly.
3.1. QUADRATICFUNCTION
The quadratic function has been a popular form of estimating crop response to
macronutrients. Possibly, because it is simple and has other attractive properties,
as described below. Its algebraic formulation is
Yi = ]~o q- /~l Ni -q- /~2 Pi q- /~3 N2 q- l~4 p2 -~ ~5 Ni Pi -q- ~i,
=l,2,...,n,
(2)
where Yi is com yield (bushels per acre), Ni is applied nitrogen (pounds per acre),
Pi, is applied phosphorus (pounds per acre), flj (j -- 1, 2 . . . . . 5) are the parameters
to be estimated, and Ei are the unexplained errors. The quadratic form imposes the
restriction of non-zero elasticity of substitution (cr ~ 0) and the absence of growth
329
DO FARMERS OVERUSE NITROGEN FERTILIZER?
plateau. 4 That is, for/31,/32 > 0 and/33,/34 < 0, yield decreases as Ni, and Pi levels
become large, ceteris paribus, so that the function exhibits diminishing marginal
productivity and input substitution for all Ni and Pi > 0.
3.2.
V O N LIEBIG FUNCTION
(3)
Y / = min(Y*,/31 +/32 Ni,/33 +/34 Pi),
where Y* is maximum corn yield and all other notations are the same as above.
The formulation is consistent with the view that nitrogen and phosphorous play
different biochemical roles in plant growth (Wild and Jones 1988). Moreover, it
implies that the plant responds in a linear fashion only to the most limiting nutrient.
After some level of application (say, N* and P* ), the plant will no longer respond
to the applied nutrients. At this point, the plant reaches maximum growth (yield)
at Y*. Thus, the von Liebig function, a priori, imposes a zero substitution elasticity
among inputs, and a growth plateau after N* and P*.
Given a wide range of agronomic nutrients, it is possible that characteristics
imposed by both the quadratic and von Liebig models are not appropriate under
all plant growth situations. For example, contrary to the von Liebig function,
macronutrients may not exhibit strictly fixed proportions for all data sets and plant
growth situations. The possibility of a > 0 and a growth plateau has considerable
intuitive appeal while modeling corn yield response to the use of nitrogen and
phosphorus.
3.3. THE MITSCHERLICH-BAULEFUNCTION
= / 3 0 [1 -
+ U,))] [1 -
+ P,))]
(4)
This function allows for plateau growth and convex, but not necessarily L-shaped,
isoquants. There are good reasons why this function is appealing. First, it can
accommodate the cases from near perfect factor substitution (a --4 cx~)to near zero
factor substitution (a --+ 0). Thus, the function is quite flexible about the extent of
isoquant convexity depending upon the data and production process. 5 Second, like
von Liebig, Mitscherlich-Baule imposes plateau growth which has considerable
appeal for many, though probably not all, plant nutrients response relationships.
Thus, in terms of isoquant pattern and plateau versus yield decline, it allows an
analyst a broader range of possibilities in model evaluation.
As reported in section 4.3, a quadratic function modified to reflect MitscherlichBaule properties gave the most satisfactory results.
4. Empirical Results
As explained above, a panel data set is used to estimate the corn response function.
Experimental data with several treatments are spread over four years, 1987 through
330
SATYA N. YADAV ET AL.
1990. In order to estimate yield function, we followed a two-step procedure. First,
possible effects o f treatment, year, and interaction between year and treatment were
examined through analysis o f variance. Second, a yield function, consistent with
the underlying theory o f an agronomic response function, was estimated based on
the inference(s) obtained from the first step.
4.1.
ANALYSIS
OF VARIANCE (ANOVA)
Corn yields for each site in a given year for a particular treatment, Yij, can be
explained as a sum o f five components: A c o m m o n value (#), an effect due to year
(r an effect due to a particular treatment (r an interaction effect o f year and
treatment ( r 1 6 2 and a residual quantity (e). This can be represented as,
i = 1, 2, 3 , . . . , r(years) and j = 1,2, 3 , . . . , t(treatments),
where
#
= overall mean, a constant,
r
Cj
= /~i. - - ~ ---- effect due to an h-th year,
= #.j - # = effect due to j-th treatment,
((~r
=
(r162 - # effect due to an interaction o f an h-th year with j-th
treatment, and
~ij
=
random c o m p o n e n t explaining all extraneous variation, and are
assumed to be independently and identically distributed normal
random variables [that is, e's are iidN (0, a~), for all ij].
From the observed data, following sum o f squares were calculated for each site. 6
C -- correction term -- Y2../rt
SSTOT -- total sum o f squares
r
t
r
t
i
j
i
j
SSYR -- sum o f squares o f years
r
= ~
i
t
1@-,
2
~ ( f . ~ _ f..)2 = ~ ,__, Y j - C
j
j
SSTRT -- sum o f squares o f treatment
r
= E
t
E(g.{
j
1
r
= -r E Y , .
{
- c
331
DO FARMERS OVERUSE NITROGEN FERTILIZER?
Table IV. Basic ANOVA table for year, tratment, and interaction effects on yield
Source of
variation
Degrees of
Sum of
freedom (DF) squares(SS)
Mean square
(MSE=SS/DF)
Expected mean square
(EMS)
Years
Treaments
Interaction
Exp. error
r-1
t-1
1
(r-1)(t-l)-I
SSYR
SSTRT
SSINTER
SSE
SSYR/(r-1)
SSTRT/(t-1)
SSINTER/I
SSE/(r-I) (t-1)-I
MSE/MSE for error
MSE/MSE for error
MSE/MSE for error
Total
(rt- 1)
SSTOT
S S I N T E R = sum o f squares o f interaction
r
t
= rt[~-~ ~
i
YG(~.-
?.)(?J
- ?.)]2/(SSYR)(SSTRT)
j
SSE -- sum o f squares o f experimental error
r
= ~
t
~-'~ (Y/j - l ? / . - l?.j + l~..) 2 - S S I N T E R
i j
= S S T O T - SSYR - S S T R T - S S I N T E R
A relationship o f the sum o f squares formulations is summarized in Table IV.
The following null hypotheses are tested with the aid o f ANOVA tables. 7
- the year effects do not exist,
there are no treatment effects, and
the interaction effects among years and treatments are irrelevant.
-
-
Table V shows ANOVA results, including inferences based on F-tests. for each
site, the effects o f year and treatments using nitrogen fertilizers were found to be
statistically significant at the 5% level, while the interaction effect o f year and
treatment was not found to be significant. A couple o f implications can be drawn
from these results. First, both years and treatments involving uses o f nitrogen
fertilizers are important variables in explaining variations in corn yields at each
site. Second, the interaction term does not have a significant effect in explaining
corn yield at any o f the three sites. As shown by Lentner and Bishop (1986), the
later finding, however, is not unexpected for this type o f experimental data.
4.2. MEAN DIFFERENCE TESTS
Site specific annual corn yields, averaged over all treatments, are given in Table VI.
At site 1, mean corn yield in 1987 was higher than in any other year. The possible
332
SATYAN. YADAVET AL.
Table V. ANOVA of year, treatment, and interaction (between year and treatment) effects for
sites 1, 2, and 3
Source
DF
SS
MSE
EMS
Inferences
3
9
1
8
1
26
15760
25887
25166
721
407
3632
5253
25166
90
407
140
42.33
179.75
0.64
2.91
-
Reject Ho
Reject Ho
Cannot reject Ho
Cannot reject Ho
-
39
45686
3
11
2
9
1
32
84234
5989
5534
455
110
4352
28078
2767
51
110
136
206.45
20.34
0.37
0.81
-
Reject Ho
Reject Ho
Cannot reject Ho
Cannot reject Ho
-
47
94685
3
15
2
13
1
44
41921
27676
24936
2740
4
3605
13974
12468
211
4
82
170.41
152.05
2.57
0.05
-
Reject Ho
Reject Ho
Cannot reject Ho
cannot reject Ho
-
63
73206
Site 1
Years
Treatments
No-N vs rest
Among N levels
Interaction
Exp. error
Toml
Site 2
Years
Treatments
No-N vs rest
Among N levels
Interaction
Exp. error
Total
Site 3
Years
Treatments
No-N vs rest
Among N levels
Interaction
Exp. error
Toml
r e a s o n s f o r l o w e r y i e l d s (in a b s o l u t e t e r m s ) in 1988 a n d 1990 c o u l d b e b e l o w n o r m a l
r a i n f a l l in 1988 a n d a b o v e n o r m a l r a i n f a l l in 1990. A d d i t i o n a l s t a t i s t i c a l t-tests,
h o w e v e r , s h o w e d that m e a n c o r n y i e l d s a c r o s s y e a r s w e r e n o t s i g n i f i c a n t l y d i f f e r e n t
f r o m e a c h Other. A t site 2, m e a n c o r n y i e l d in 1988 w a s f o u n d to b e s i g n i f i c a n t l y
d i f f e r e n t a n d the l o w e s t a m o n g all y e a r s . T h e t-tests, n e v e r t h e l e s s , s h o w e d that
m e a n c o r n y i e l d s f o r y e a r s 1987, 1989, a n d 1990 w e r e n o t s i g n i f i c a n t l y d i f f e r e n t
f r o m e a c h other. F i n d i n g s o f m e a n c o r n y i e l d d i f f e r e n c e s f o r site 3 w e r e s i m i l a r to
site 2. T h e y e a r 1988 h a d the l o w e s t y i e l d , w h i l e y e a r s 1987, 1989, a n d 1990 h a d
similar corn yields.
T r e a t m e n t - w i s e c o r n y i e l d s , a v e r a g e d o v e r all f o u r y e a r s , a r e p r e s e n t e d in T a b l e
V I I . A s e x p e c t e d , m e a n c o r n y i e l d f o r z e r o l e v e l o f n i t r o g e n ( t r e a t m e n t 1) w a s
f o u n d to b e s i g n i f i c a n t l y l o w e r t h a n all n o n - z e r o l e v e l s o f n i t r o g e n at site 1. H o w ever, m e a n c o r n y i e l d s a m o n g t r e a t m e n t s i n v o l v i n g n o n - z e r o l e v e l s o f n i t r o g e n
333
DO FARMERS OVERUSE NITROGEN FERTILIZER?
Table VI. Annual mean corn yields (bu/acre) for the three sites by year
Years
1987
1988
1989
1990
Average
Site l
Comyields (bu/acre)
Site2
Site3
187.12
(29.1)
146.85
(25.2)
166.49
(35.2)
134.89
(26.5)
188.85
(8.07)
80.340
(14.5)
165.92
(16.7)
170.31
(26.0)
177.71
(22.5)
109.04
(18.7)
125.09
(32.2)
130.60
(20.7)
158.84
(35.15)
151.35
(45.46)
135.61
(35.05)
Note: The computations are based on observations from replications. Sample sizes
are Site 1 = 160 (each year 40), Site 2 = 192 (each year 48), and Site 3 = 256 (each
year 64). Figures in parentheses are standard deviations of the means.
applications, surprisingly, were not statistically different from each other at 1%
significance level (Table V). A considerable amount of residual nitrogen build-up
(over several years) in the soil profile could be one possible reason. If so, such
accumulated nitrogen should be considered as a valuable source of nitrogen supply for the plant. Also, corn yields among treatments involving the recommended
dose of nitrogen, 150 lb/acre, were not statistically different from each other, irrespective of variations in source, time, season, use of nitrogen inhibitor, method of
fertilizer application, and tillage practices used. These later findings have important
implications. More environmentally favorable cultural practices could be chosen
for nitrogen application without sacrificing profits.
Results for site 2 and site 3 were similar to site 1. Mean corn yields for zero
levels of nitrogen were significantly (at 1% level) lower than the non-zero levels
at both sites 2 and 3. However, mean yields did not differ significantly among
treatments using non-zero levels of nitrogen.
4.3. YIELD RESPONSE FUNCTION
Characteristics of agronomic response functions, discussed above, prompted us
to select the Mitscherlich-Baule functional form. Several attempts were made to
estimate the Mitscherlich-Baule function using non-linear estimation packages.
However, the direct estimation of this function did not provide a good fit. In
contrast, the quadratic form gave a good fit to the data and a separate function was
estimated for each site. 8 However, the estimated quadratic function was modified
to reflect the properties of the Mitscherlich-Baule function. The values of N* were
obtained by maximizing the respective function to serve as plateau for N in the
334
SATYA N. YADAV ET AL.
Table VII. Treatment-wise mean corn yields at the three sites
Treatments
Site 1
No. 1
Corn yields (bu/acre)
Site 2
No. 11
83.58
(24.5)
155.50
(19.0)
172.56
(23.0)
167.36
(13.8)
169.45
(22.3)
169.13
(19.9)
168.47
(19.1)
167.96
(24.2)
165.92
(38.5)
168.44
(34.5)
-
No. 12
-
No. 13
-
126.72
(43.8)
153.10
(48.1)
155.76
(45.1)
157.75
(45.1)
163.03
(40.4)
127.97
(46.5)
155.32
(46.0)
151.86
(50.1)
155.84
(42.8)
156.67
(43.2)
160.24
(45.4)
151.99
(47.2)
-
No. 14
-
-
No. 15
-
-
No. 16
-
-
Average
158.84
(35.15)
151.35
(45.46)
No. 2
No. 3
No. 4
No. 5
No. 6
No. 7
No. 8
No. 9
No. 10
Site 3
89.42
(27.5)
126.61
(27.5)
143.40
(28.3)
146.87
(27.4)
147.53
(28.5)
77.35
(31.8)
137.21
(31.5)
141.46
(33.5)
139.97
(32.6)
142.33
(26.2)
145.57
(26.4)
148.18
(26.1)
140.06
(30.9)
146.17
(27.7)
150.30
(29.8)
145.36
(24.3)
135.61
(35.05)
estimated functions (Table VII). Each function is concave and increasing in nitrogen
before the plateau is reached. Inferences obtained from the mean difference tests
were used to formulate appropriate dummies. As far as we know, this is the first
time such a modification of the quadratic form has been used.
DO FARMERSOVERUSENITROGENFERTILIZER?
335
Table VIII. Corn response functions for sites 1, 2, and 3
Site
Estimated corn response function
Site 1
Y = 111.87 + 0.467 N* - 0.00128 N2. - 40.28 YEAR88 - 20.63 YEAR89
(27.88) (3.14)
(-2.79)
(-12.80)
(-6.55)
-52.23 YEAR90 + 44.07 FERT
(-16.59)
(3.61)
where: N* =N, ifN <__182
N* = 182, if N > 182
Site 2
Adjusted R 2 = 0,82
Sample size = 160
Y = 167.12 + 0.368 N* - 0.00117 N 2. - 108.5 Y E A R 8 8 - 22.92 Y E A R 8 9
(56.62)
(7.20)
(-4.67)
(-37.56)
(-7.94)
-18.54 YEAR90
(--6.42)
where: N* =N, ifN _< 157
N* = 157, if N > 157
Site 3
Adjusted R 2 = 0.90
Sample size = 192
Y = 125.49 + 0.476 N* - 0.00149 N2. - 68.67 YEAR88 - 52.62 YEAR89
(46.95) (4.21)
(-3.48)
(-30.13)
(-23.08)
-47.11 YEAR90 + 24.85 FERT
(-20.67)
(3.29)
where: N* =N, ifN _< 160
N* = 160, if N > 160
Adjusted R 2 = 0.86
Sample size = 256
Note: Figures in parentheses are t-values of the corresponding estimates.
Where, Y = corn yield (bu/acre), N = nitrogen (lb/acre), YEAR.. = corresponding year
dummies (1 for given year; otherwise 0), FERT = fertilizer dummy (1 if any amount of
nitrogen applied; otherwise 0).
Table I X s h o w s the p r o f i t - m a x i m i z i n g levels o f nitrogen use, m e a s u r e d in
p o u n d s per acre. U s i n g the results o f estimated corn r e s p o n s e functions f r o m
Table VIII, p r o f i t - m a x i m i z i n g levels o f nitrogen w e r e d e t e r m i n e d f o r e a c h site (by
e q u a t i n g the c o r r e s p o n d i n g v a l u e o f marginal p r o d u c t to the price o f nitrogen) and
are s h o w n in the s e c o n d c o l u m n o f Table IX. 9 A corn price o f $ 2 . 4 0 / b u and nitrogen
price o f $0.15/lb w e r e used in the c o m p u t a t i o n . N o t e that these r e c o m m e n d e d
values do not take into a c c o u n t o f the residual nitrogen in the soil profile. In the
a b s e n c e o f further information, these are the values that w o u l d be r e c o m m e n d e d
b y e c o n o m i s t s . H o w e v e r , o u r results s h o w that such a r e c o m m e n d a t i o n w o u l d not
o n l y be m i s l e a d i n g but also detrimental to the e n v i r o n m e n t .
T h e third and the fourth c o l u m n s o f Table I X s h o w r e c o m m e n d e d levels o f
nitrogen for each site w h e n residual nitrogen available in the soil profile is taken
336
SATYAN. YADAVET AL.
Table IX. Profit-maximizing levels of nitrogen (lb/acre) for sites 1, 2, and 3a
Site
No consideration of
Consideration of residual nitrate at
residual nitrogen
0-2 feet depth
0-3 feet depth
Si~ 1
S i ~ 2b
158
131
131
00
127
00
Si~ 3
139
99
84
a Based on our conversations with the agronomists, corn roots can go down to 4
feet depth. However, most of the roots are distributed in the upper 2 feet depth
of the soil profile.
b No additional nitrogen is recommended for site 2 for the first year because of
the residual nitrogen.
into account, l~ We have considered two possible scenarios of root zone depths,
0-2 and 0-3 feet depth. In each case, recommended nitrogen application is much
lower than the case involving no consideration of residual nitrogen. When nitrogen
available in the upper 2 feet depth of the soil profile is included, the recommended
rate (column 2) can be reduced by 17% (site 1) to 100% (site 2). 11 Such a reduction
does not seem to reduce the corn yield and is supported by our findings in the
earlier section. As shown in the fourth column, reductions in nitrogen would be
still higher if the upper 3 feet depth of the soil profile is considered. Site 2 has
the highest level of stored nitrogen in the soil profile. This finding is consistent
with our previous findings of statistically no mean yield differences across zero
and non-zero level of nitrogen applications. Of course, as the residual nitrogen is
depleted, the recommended levels would increase over time, although not to the
point where there is a residual build-up in the soil.
The results also indicate that the profit-maximizing level of nitrogen use is
not same for all sites. The implication is that nitrogen recommendation should
be site specific rather than a general figure for the region as a whole. Moreover,
considering residual nitrogen as a potential source of nutrient supply for the plant,
the recommended rate should not be independent of existing nitrogen supply from
the soil profile.
The current recommended rate of nitrogen application for southeastern Minnesota is 150 lb/acre. However, farmers have been applying about 176 lb of nitrogen
per acre in this region (Wall et al. 1989). This is higher than both the recommended and profit-maximizing levels of use. 12 It appears from this study that farmers
in southeastern Minnesota could decrease nitrogen use and increase their profits from savings in fertilizer use. 13 Since farmers in this area are not atypical of
farmers throughout the country, one suspects a general overuse of nitrogen fertilizers. Decreasing the use of this input would also benefit society. Beside surface
contamination of water resources, there is a positive relationship between surface
applications of nitrogen and residual nitrogen build-up in the soil profile (Table III).
The excess nitrogen has the potential of leaching into the ground water. Reduced
DO FARMERS OVERUSE NITROGEN FERTILIZER?
337
nitrogen application by farmers thus would have a direct and positive impact on
decreasing nitrogen loading to water resources as well, thereby improving water
quality of the region. This is a win-win situation. This is not to say that the profit
-maximizing rate is optimum for society; leaching can still occur.
5. Concluding Remarks
In this study, we estimate agronomic corn response functions utilizing panel data
from three experimental sites in southeastern Minnesota. The estimated quadratic
functions are further modified to reflect the properties of concavity until the maximum is attained and a flat plateau thereafter. Optimal uses of nitrogen are derived
in the static context based on these estimated functions. The following conclusions
can be drawn from our results.
First, both the current recommendation rate, 150 lb/acre, and farmers' use,
176 lb/acre, of nitrogen in the study areas exceed the profit maximizing level
of nitrogen for the region. The current recommended rate needs to be revised
downward. Moreover, the study shows that the recommendation should be made
site or area specific rather than a general figure for the entire region.
Second, the current study shows a considerable residual nitrogen build-up in
the soil profile, implying that farmers have been applying more nitrogen than could
be utilized by corn plants. Since such nitrogen, especially that available in the
upper 2 or 3 feet of the root zone, can also be utilized by plants, appropriate credit
should be given for this nitrogen (to avoid over application), when making nitrogen
recommendations. This will not only reduce production costs of farmers but will
also reduce further deterioration of water quality in the region.
The results of this study should be taken as preliminary. Although the three sites
are not atypical of southeastern Minnesota, variations within and among regions
exist. Additional studies are needed to substantiate or refute the findings of the
study. Moreover, it should be emphasized that negative externalities from fertilizer
usage have not been considered in assessing its profitability. Once such social costs
are included in the analysis, the socially optimum level of nitrogen use would still
be lower.
Our study has also identified several areas for further research. First, since
residual nitrogen in the soil profile and the level of contamination vary across the
state, it is essential to have similar studies done for the other parts of the state,
especially where nitrate pollution is perceived to be a serious problem.
Second, this study shows that farmers apply more nitrogen than required to
attain the same level of corn yields. The study, however, does not explain the
reason behind it. An extensive study of why farmers use too much nitrogen could
provide invaluable insights into how nitrogen application rate can be reduced. Also,
the study focus entirely on continuous corn. Thus similar studies should be done
for the other common crop rotations in the state.
338
SATYAN. YADAVET AL.
Third, this study deals with the continuous com where fertilization is primarily
through chemical fertilizers. There are many corn farms, however, that use different combinations of chemical fertilizers and manure. A comprehensive study
combining both sources of nitrogen appears to be important for making nitrogen
recommendation for these farms.
Fourth, as pointed out in the text, agriculture (though major) is one of the sources
of nitrate pollution. An extensive study on contribution of various other sources
of nitrate pollution could be helpful in further curtailing pollution. Moreover, such
steps would help assure farmers that they are not the only ones reducing their
nitrogen use.
Notes
1. An externality is said to exist when individual X's utility or production function consists of
nonmonetary variables, values of which are chosen by others without giving any regard to its
effects on X's welfare. Moreover, the decision maker does not receive (or pay) in compensation
for this activity an amount equal in value to the resulting benefits (or costs) to individual X
(Baumol and Oates 1988).
2. EPA has set a maximum contamination level of 10 mg/l nitrate-nitrogen (NO3-N) or 45 mg/l of
NO3, to be safe for drinking water.
3. Size of each plot was 65 feet long and 30 feet wide.
4. Elasticity of substitution ((r) of P for N is defined as,
d(Pi/N~) / d(fN/fP)
where the differentials are restricted to variations along an isoquant. Thus, # refers only to input
substitutions associated with a constant level of output. An expression for a based upon the
quadratic function is
0"~
--fN fp(N, fN + P~ fp)
N, P,(f.N f~. -- 2 f . ~ fN fv + f l . . f~)
which is non-zero (Ferguson 1975).
5. The elasticity of factor substitution at the h-th data point for the Mitscherlich-Baule function is
Y~[fN Ni + fpP,]
N~ P~[2fN fp + I~3Y~ f2v +/~1 ~ fP]
where fN = ~o ~l [1--exp(--fla (/34+Pi))] exp(-~l (~2+Ni))] and fp =/~o/33 I1-exp(-fll (/~2+
Ni))] exp(-/~3(fl4 + PI))], are the first derivatives evaluated at Ni and Pi respectively. The
above result is obtained by substituting the values of fray = -~1 fN, fro, = f13 fv, and
fNP = f~v f p / Y into the usual formula for ~r. The isoquants approach the limiting case of
no factor substitution (right angle isoquants) as Ni and Pi get arbitrarily large. In contrast, ai
increases as N~ and Pi get smaller, thereby suggesting a greater factor substitution.
6. The approach followed is from Lentner and Bishop (1986).
7. The data set consist of observations on the dependent variable for all categories of the explanatory
variables. Therefore, it satisfies the assumptions of a fixed effect model. Accordingly, inferences
discussed here are for the fixed effects model. See Iversen and Norpoth (1987) and Lentuer and
Bishop (1986) for detailed discussion on this issue.
8. Characteristics of treatments vary across sites. Therefore, a single response function from pooled
data would not be an appropriate procedure.
DO FARMERS OVERUSE NITROGEN FERTILIZER?
339
9. Note that negative externalities from fertilizer uses have not been considered in assessing its
profitability. Once such social costs are included in the analysis, the socially optimum level of
nitrogen use would be still lower. See Yadav (1994) for a detail treatment on this issue.
10. We believe that although left-over nitrogen (after plant uptake) cannot be removed from the soil
profile, it can be utilized by the successive crops. Following the approach of Rehm and Schmitt
(1990), we give full credit to residual nitrate-nitrogen in our analysis.
11. The recommended levels of nitrogen under the third and the fourth columns have been derived by
subtracting the average residual nitrogen available in the corresponding soil profile depth from
Table II. The residual nitrogen consisted of nitrogen available in nitrate form. We used the data
collected in April, 1987 (i.e., prior to the start of experiment) so as to reflect the existing nitrogen
build-up in the soil profile in southeastern Minnesota.
12. Possible explanations for the high application rates include: overestimation of yield goals,
perceiving fertilizer as a cheap insurance against nitrogen losses in the root zone, and discounting of the nutrient content of manure. Also, farmers may treat fertilizer as a risk-reducing
input (Olson 1986).
13. This is supported by the fact that farmers in one of the study areas were able to reduce nitrogen use
in corn by 21%. The reduction took place under the Rural Clean Water Program on a voluntary
basis and did not reduce crop yield (Wall et al. 1989).
References
Baumol, W. J. and W. E. Oates (1988), The Theory of Environmental Policy.Cambridge, New York:
Cambridge University Press.
De Janvry, A. (1972), 'The Generalized Power Function', American Journal of Agricultural Economics 54, 234-237.
Frank, M. D., B. R. Beattie, and M. E. Embleton (1990), 'A Comparison of Alternative Crop Response
Models', American Journal of Agricultural Economics 72, 597-603.
Ferguson, C. E. (1975), The Neo Classical Theory of Production and Distribution. Cambridge, New
York: Cambridge University Press.
Griffin, R. C., J. M. Montgomery, and M. E. Rister (1987), 'Selecting Functional Form in Production
Function Analysis', Western Journal of Agricultural Economics 12, 216-227.
Grow, S. R. (1986), Water Quality in the Forestville Creek Karst Basin of Southeastern Minnesota.
Unpublished M.S. thesis, University of Minnesota.
Hallberg, G., R. D. Libra, and B. E. Hoyer (1985), Nonpoint Source Contamination of Groundwater
in Karst Carbonate Aquifers in Iowa, Perspectives on Nonpoint Source Pollution, Proceedings
of a National Conference, U.S. Environmental Protection Agency, Washington, DC.
Heady, E. O. (1952), Economics of Agricultural Production and Resource Use. Englewood Cliffs,
New Jersey: Prentice-Hall.
Heady, E. O. and J. L. Dillon (1961), Agricultural Production Functions. Ames, Iowa: Iowa State
University Press.
Heady, E. O. and R. W. Hexem (1978), Water Production Functionsfor Irrigated Agriculture. Ames,
Iowa: State University Press.
Iversen, G. R. and H. Norpoth (l 987), Analysis of Variance. Second edition, Newbury Park, California:
Sage Publications.
Jannik, N. O., E. C. Alexander, Jr., and L. J. Landher (1991), The Sinkhole Collapse of the Lewiston,
Minnesota Waste Water Treatment Facility Lagoon, Presented at the Proceedings of the Third
Conference on Hydrogeology, Ecology, Monitoring, and Management of Groundwater in Karst
Terrains, Clarion, Nashville, Tennessee, December.
Johnson, P. R. (1953), 'Alternative Functions for Analyzing a Fertilizer Yield Relationship', Journal
of Farm Economics 35, 519-529.
Johnson, S. L., G. M. Perry, and R. M. Adams (1989), Managing Groundwater Pollution From
Agricultural Related Sources: An Economic Analysis, Final Technical Report Submitted to U.S.
Geological Survey, Renton, VA by Dept. of Agricultural & Resource Economics, Oregon State
University, Corvallis, OR.
340
SATYAN. YADAVET AL.
Lee, L. K. and E. G. Nielsen (1987), 'The Extent and Costs of Ground Water Contamination by
Agriculture', Journal of Soil and Water Conservation 42, 243-248.
Lentner, M. and T. Bishop (1986), Experimental Design and Analysis. Blacksburg, Virginia: Valley
Book Company.
Minnesota Department of Agriculture (1991), Minnesota Agricultural Statistics, P.O. Box 7068, 90
West Boulevard, St. Paul, MN 55107.
Minnesota Pollution Control Agency (1989), Minnesota Ground Water Protection Strategy, September.
Olson, R. A. (1986), Agricultural Practices for Minimizing Nitrate Content of Ground Water, in
Proceedings of Agricultural Impacts of Ground Water: A Conference, Omaha, Nebraska, August
11-13.
Perrin, R. K. (1976), 'The Value of Information and the Value of Theoretical Crop Response Research',
American Journal of Agricultaral Economics 58, 54-61.
Pye, V. I. (1983), Groundwater Contamination in the United States, Workshop on Groundwater
Resources and Contamination in the United States (Summary and Papers), National Science
Foundation, Washington, D.C.
Randall, G. W., J. L. Anderson, G. L. Malzer, and B. W. Anderson (1990), Impact of Nitrogen
and Tillage Management Practices on Corn Production and Potential Nitrate Contamination
of Ground Water in Southeastern Minnesota, Center for Agricultural Impacts on Water Quality,
University of Minnesota.
Rehm, G. and M. Schmitt. (1990), Fertilizer Recommendations for Agronomic Crops in Minnesota,
Minnesota Extension Service, Report # AG-MI-3901, University of Minnesota.
Spillman, W. J. (1923), 'Application of the Law of Diminishing Returns to Some Fertilizer and Feed
Data', Journal of Farm Economics 5, 36-52.
U.S. Department of Agriculture (1991), Agricultural Chemical Usage 1990: Field Crops Summary,
National Agricultural Statistics Service, Economic Research Service, Washington, DC.
Vroomen, H. (1989), Fertilizer Use and Price Statistics, Resource and Technology Division, ERS,
USDA, Statistics Bulletin # 780, p. 20.
Wall, D., S. A. McGuire, and J. A. Magner (1989), Water Quality Monitoring and Assessment in the
Garvin Brook Rural Clean Water Project Area, Division of Water Quality, Minnesota Pollution
Control Agency, St. Paul, Minnesota.
Wall, D. ( 1991 ), 'Nitrate in Groundwater- Existing Conditions and Trends', in Nitrogen in Minnesota
Groundwater, Prepared for the Legislative Water Commission, December.
Wild, A. and L. H. P. Jones (1988), 'Minerzal Nutrition of Crop Plants', Chapter 3, in A. Wild, ed.,
Russell's Soil Condition and Plant Growth. New York: John Wiley and Sons.
Yadav, S. N. (1994), Nitrate Contamination of Ground Water in Southeastern Minnesota: A Dynamic
Model of Nitrogen Use, Unpublished Ph.D. dissertation, Department of Agricultural and Applied
Economics, University of Minnesota.