




































In ternationa l
Scholars
Journa ls

 

African Journal of Agricultural Marketing ISSN 2375-1061 Vol. 7 (2), pp. 001-007, February, 2019. Available online at 
www.internationalscholarsjournals.org © International Scholars Journals 

 

Author(s)                                           Author(s) retain the copyright of this article. 
 
 

 

Full Length Research Paper 

 

Analyzing the sources of technical efficiency 
among heterogeneous dairy farms: A quantile 

regression approach 

 
Benaissa Chidmi1*, Daniel Solís2 and Victor E. Cabrera3

 
 

1
Department of Agricultural and Applied Economics, Texas Tech University, Lubbock 79409, Texas, USA. 

2
Division of Marine Affairs and Policy, RSMAS, University of Miami, Miami 33149, Florida, USA. 

3
Department of Dairy Science, University of Wisconsin, Madison 53706 Wisconsin, USA. 

 
Accepted 13 November, 2018 

 
An unbalanced panel data including 1,151 farm observations from 2004 to 2008 was used to analyze the 
determinants of technical efficiency (TE) for dairy farms in the State of Wisconsin. To account for farm 
heterogeneity in our analysis we implemented a two-step framework using a stochastic production 
frontier and a quantile regression analysis. The results show that the determinants of TE affected 
farmers with different levels of TE in a very specific ways. This outcome is of significant importance 
from an empirical point of view. Farmers could use this knowledge to find alternatives to improve their 
specific level of performance. Additionally, Policy makers could use this information to improve the 
effectiveness of their policies by targeting specific agricultural services and aid to group of farmers 
with similar levels of TE. 

 
Key words: Dairy farm, technical efficiency, quantile regression approach. 

 
INTRODUCTION 

 
The US dairy industry is facing several economic 
challenges and opportunities at both the international and 
domestic levels. At the international level, the Uruguay 
Round of the General Agreements on Tariffs and Trade 
imposes limits on the use of subsidized exports and also 
transforms dairy import quotas into tariffs. On the other 
hand, increasing demand for dairy products from 
developing as well as from developed countries offers 
viable opportunities for this industry (Murova and Chidmi, 
2009). At the domestic level, dairy markets are been 
shaped by several factors including: 1) structural changes 
in the dairy industry (for example, large size and smaller 
number of dairy farms, consolidation of dairy coopera-
tives, and consolidation of retailers), 2) the dynamics of 
consumer demands, and 3) changing policies. 
Additionally, dairy farms in traditional states must  
 
 

 
*Corresponding author. E-mail: benaissa.chidmi@ttu.edu Tel: 1 
(806) 742-1921, Ext. 250. 

 
 
 

 
compete against an ever growing supply from emerging 
dairy states (Cabrera et al., 2008). Under these circum-
stances, several studies have suggested that dairy farms 
in traditional production areas must improve their levels of 
technical efficiency (TE) if they are to survive in this 
complex and evolving market (Tauer, 2001; Alvarez et al., 
2008; Cabrera et al., 2010).  

The accurate analysis of the determinant of TE is 
critical to the dairy farmers as well as to policy makers. 
For the farmers, understanding how different factors 
affect their TE is a helpful tool for improving the 
performance and profitability of their dairy farms. From 
the policy makers‟ viewpoint, knowing the distribution of 
TE across dairy farms will help to draft specific and well 
defined dairy policies, which would increase TE and the 
competitiveness of this industry.  

Consequently, the objective of this study is to evaluate 
the determinants of TE among dairy farms in the State of 
Wisconsin. Special attention is given in this work to 
account for the potential effect of farm heterogeneity 
when analyzing the sources of farm inefficiency. Previous 

file:///C:\Users\user\Documents\REPUBLICATION\AGRICULTURAL%20SCIENCES\AppData\Local\Temp\www.internationalscholarsjournals.org


 
 
 

 

studies addressing the issue of farm heterogeneity have 
either used „expert-knowledge‟ (For example, Newman 
and Matthews, 2006; Tauer, 2005; Bravo-Ureta, 1986) or 
statistical techniques (Alvarez and del Corral, 2010; 
Alvarez et al., 2009) to divide the sample using some 
specific technological characteristics (that is, herd breed, 
milking systems, specialization) and then estimate 
independent production frontiers for groups of farmers 
with different technologies. However, Cabrera et al. 
(2010) show that for a cross-section of the sample used 
in this study, technological differences (that is, milking 
systems, housing, and the use of pastures) have no 
significant impact on the farm level of efficiency. Similar 
outcomes can be found in Tauer (2006), Hallan and 
Machado (1996), and Bewley et al. (2001). Thus, in this 
study an alternative framework to study the sources of 
farm inefficiency while controlling for farm heterogeneity 
was proposed. Specifically, we implement a two-step 
framework, in which the TE scores are computed using 
the stochastic production frontier (SPF) method and then 
determinants of inefficiency are analyzed using a quantile 
regression. This framework will allow us to assess the 
specific impact of different farm‟s managerial 
characteristics on the TE of alternative groups of farms 
clustered on their own level of efficiency. The running 
hypothesis is that impact of different factors on the dairy 
farm TE will vary depending on how far each farm is from 
the production frontier.  

It is important to indicate that previous studies using a 
two-step approach based on Ordinary Least Square 
and/or dichotomous (Probit and Tobit) regressions have 
been criticized due to inconsistencies in the distribution of 
the TE score and the distribution assumed in the second 
step. In this study, we control this issue by using quantile 
regression that offers the flexibility for modeling data with 
heterogeneous conditional distributions and makes no 
distributional assumption about the error term in the 
model (Chen, 2005; Koenker and Hallock, 2001).  

This paper also adds to the literature by analyzing the 
impact on TE of some variables which have not received 
much attention in the past such as government 
payments, alternative sources of farm income and farm 
financial health. 

 

MODEL 
 
The study implements a two-step approach to analyze the level and 
determinants of TE among a sample of dairy farms in Wisconsin. In 
the first step, the study estimates an SPF following the framework 
proposed by Aigner et al. (1977). The SPF method is based on an 
econometric (that is, parametric) specification of a production 
frontier. Using a Cobb-Douglas production function and panel data, 
this method can be presented as: 
 

m  
ln

 

y
it   

=
 ∑β

i  

lnx
ijt  
−

 

ε
it (1) 

i0 
 
Where: y represents output, x  is the amount of inputs, βs  are the 

  
  

 
 

 
unknown parameters, and ε is the error-term. The subscripts i, j, and t 

denote the farm, inputs and time, respectively. The error-term is farm-  

specific and is composed of two independent components, ε it   = v it  -  

u it  . The first element, v, is a random variable reflecting noise and 
 
other stochastic shocks which is assumed to be an independent 
and identically distributed normal random variable with 0 mean and 

constant variance, iid [N~(0,σv
2
)]. The second component, u, 

captures technical inefficiency (TI) relative to the stochastic frontier. 
The inefficiency term u is non-negative and it is assumed to follow a 
half-normal distribution (Kumbhakar and Lovell, 2000).  

An index for TE can be defined as the ratio of the observed 
output (y) and maximum feasible output (y*): 
 

TEi= 
y 

i   
f(xij ;β)  expvi  − ui  

 exp− ui ;  TI  1− TE       

yi 
  

f(xij ;β)  expvi  
 

  *  
 

       (2) 
 

 
In the second step, the study uses quantile regression to regress TI 

on a set of variables, z, that influence the inefficiency term ui: E(TE 
| Z  z)  z'θ. The conditional quantile parameters can be 
 
 n   

 estimated by solving
 θ  argmin∑ρτ (TIi  − z'i θ ) 

, with
  ρτ    τ  

if
 

 

 i1   
 

the observation belongs to the τ 
th

  quantile and  ρ 
τ  1 − τ  if not. 

 

    

 

DATA AND EMPIRICAL MODEL 

 
The data used in this study consisted of detailed farm-level 
information for dairy farms participating in the Agriculture Financial 
Advisor (AgFA) program managed by the Center for Dairy 
Profitability at the University of Wisconsin-Madison. The empirical 
sample included 1,151 Wisconsin dairy farm observations and the 
collected information covered the period of 2004 to 2008. The dairy 
farms in the sample were highly specialized with most of their 
output coming from dairy sales. It is important to indicate that 
Cabrera et al. (2010) used a cross section of this data set (2007 
agricultural year) to analyze the determinants of TE, using a one-
step SPF framework following Caudill et al. (1995).  

As indicated, the first-step in the empirical analysis is based on 
the estimation of a Cobb-Douglas production function. The 
dependent variable is the total milk production sold and measured 
in kg. Following Cabrera et al. (2010), 6 inputs were included in this 
study: „cow‟, defined as the number of adult cows (all cows after 
first calving) in the herd and which measures the livestock capital; 
„feed‟, defined as the total cost of purchased feedstuffs in US $; 
„capital‟, defined as the depreciation of buildings and land, 
corresponding to 5% of the value of land used by the farm; „crop‟, 
defined as the total expenses related to crop production measured 
in US $ (including chemicals, fertilizers, lime, seeds and plant 
purchases, machinery depreciation, machinery hire expenses, 
machinery repair, fuel and oil expenses); and „labor‟, defined as the 
total labor including family and hired labor measured in US $. In 
addition, a dummy variable for the year 2005 (drought) was 
included to account for a severe drought that affected the 
production of grass and other agricultural products. For the fact that 
the sample is highly unbalanced (that is, farms vary between time 
period), it is inefficient to implement any panel data techniques 
(fixed and random effects) in this study. To alleviate the problem, 
this study included the regression of a time trend variable (t) to 
account for any technical change during this period.  

In the second-step, TI was regressed on several managerial 
characteristics of the dairy farms. The inefficiency model included a 



 
 
 

 
Table 1. Descriptive statistics for Wisconsin dairy farms (N=1,151, yrs. 2004 to 2008).  

 
 Variable (unit) Mean Std Dev Minimum Maximum 

 Milk (kg) 3,081,791 4,212,449 360,807 47,972,970 

 Cows (n) 140 176 23 1,844 

 Feed ($) 116,023 184,823 2,760 1,867,926 

 Capital ($) 90,905 97,070 6,279 1,626,164 

 Crop ($) 139,325 151,217 3,733 1,585,638 

 Labor ($) 57,188 106,025 161 1,227,002 

 TMR (dummy)
1
 0.53 0.50 0 1 

 Pature (dummy)
2
 0.16 0.37 0 1 

 Milking system (dummy)
3
     

 Flat barn 0.09 0.28 0 1 

 Pit parlor 0.26 0.44 0 1 

 Milking frequency (dummy)
4
 0.91 0.29 0 1 

 bST (%) 14.62 25.07 0 100 

 Family labor (%) 58.96 44.09 0 100 

 Feed/cow (ratio) 705.50 325.98 52.77 2,026.65 

 Housing (dummy)
5
 0.39 0.49 0 1 

 Government payments ($) 0.16 0.16 0.00 1.06 

 Nonfarm income ($) 0.14 0.27 0.00 3.12 

 Calves sold ($) 0.12 0.26 0.00 3.90 

 Crop sold ($) 0.18 0.38 0.00 3.19 

 Family savings ($) 0.49 0.50 -2.02 3.26 

 Investment/cow (ratio) 0.12 0.05 0.04 0.38 

 Debt/cow (ratio) 0.03 0.02 0.00 0.11 
 

1
Use of TMR = 1, 

2
Use of pasture = 1, 

3
Pipeline is the omitted variable, 

4
Two times daily milking frequency = 1, 

5
Free stall housing = 1. 

 

 
set of milking system dummy variables, like: flat barn and pit parlor 
(pipeline was the omitted variable); milking frequency, which is a 
dummy variable that equals 1 for the farms with a milking frequency 
equal to 2 (0 equals more than 2 times); bST, which is the percent 
of cows under bovine somatotropin treatment; family labor, which is 
the ratio of family labor over total labor; and housing, which is a 
dummy variable that equals 1 for farms that use free stall housing. 
During the last years, several studies have shown the importance of 
intensification of production on the efficiency of dairy farms 
(Ledgard et al., 2004; Alvarez et al., 2009; Cabrera et al., 2010). 
Thus, to assess the impact of intensification on efficiency, the study 
included the following variables: feed/cow, defined as the ratio of 
purchased feedstuffs to the number of cows; TMR, a dummy 
variable equal to 1 for the farm that used the „total mixed ration‟ 
feeding system; and pasture, a dummy variable equal to 1 for farms 
that used pasture feeding systems. Additionally, the study also 
analyzed the effect of other sources of income on TE by including 
the variables of government payments, non-farm income, and the 
revenue from calves and crops sold, all measured in US $. Finally, 
to study the effect of the dairy farm‟s financial health on efficiency, 
the family savings and investment per cow were included, as well 
as a ratio measuring the debt per cow. Table 1 presents descriptive 
statistics for all the variables included in the analysis. 

 

RESULTS AND DISCUSSION 

 

Frontier analysis 
 

Table 2 presents the maximum likelihood estimates of the 

 
 

 

Cobb-Douglas production frontier model from the first-
step. All estimated parameters are positive and, with the 
exception of capital, they are all statistically significant. 
Given that all input variables and the output are in 
logarithmic form, the parameter estimates can be 
interpreted as partial production elasticities. The empirical 
results indicate that the variable that contributes the most 
to farm production is „cows‟. Specifically, a 10% increase 
in the number of cows in the herd translates in an 
increase in milk production sold of 7.78%. The next 
highest elasticity is for „feed‟ (1.34%), followed by „crops‟ 
(0.71%), and „labor‟ (0.45%).  

The time trend (t) is negative and statistically significant 
indicating a decreasing rate in production levels during 
the studied period. This result agrees with the findings 
presented by Ball (2009) who shows that during the last 
decade traditional dairy states in the US, including 
Wisconsin, have decreased their level of production, 
while states in the American West and Southwest have 
displayed significant improvements. In addition, the 
parameter for the variable drought is also negative and 
significant suggesting that the adverse climatic condition 
of 2005 negatively affected the farms‟ average level of 
production. This outcome confirms the idea stated by 
Demir and Mahmud (2002) regarding the importance of 
controlling the climatic and environmental conditions 



  
 
 

 
Table 2. Production frontier estimates (N=1,151, yrs. 2004 to 2008). 

 

 Variables
1
 Coefficient St. Dev. t-value 

 Constant 8.489*** 0.099 85.22 

 Cow (n) 0.778*** 0.017 46.23 

 Feed ($) 0.134*** 0.008 17.56 

 Capital ($) -0.006 0.010 -0.60 

 Crop ($) 0.071*** 0.009 7.44 

 Labor ($) 0.045*** 0.004 11.72 

 Time Trend -0.019*** 0.003 -6.18 

 Drought (dummy) -0.034*** 0.010 -3.35 

 σv 0.066*** 0.005 12.67 

 σu 0.199*** 0.008 2489 

 Log-likelihood 701.98   

 Log-likelihood ratio 0.90 Fail to reject H0 
 

*P < 0.10; **P < 0.05; ***P < 0.01. 
1
Dependent variable is the total milk production sold measured in kg. 

 

 

when studying TE in agriculture.  
The empirical results also suggest the presence of 

constant returns to scale (CRS). Specifically, the scale 
elasticity (that is, the sum of all output elasticities) is 
equal to 1.022. This outcome is confirmed by a likelihood 
ratio test that failed to reject the hypothesis 

H 0  : ∑β i    1 . Kompas and Chu (2006) indicates that 
 
CRS implies that, for the studied sample, productivity 
depends on improvements in technology and efficiency, 
and not necessarily on the size of the farm. The 
implication of CRS is that there is no scale effect in the 
size of the farm: the output produced and the farm size 
will be proportional. Therefore, an improvement in 
productivity (not production) can only come from 
improvement in technology and efficiency and not from 
the farm size. However, the size will affect the production 
(not the productivity). Given constant returns to scale, the 
dairy farms in our sample are scale efficient.  

The distribution of the TE estimates is presented in 
Figure 1. The results indicate that, on average, the 
studied dairy farms have a TE exceeding 90%, with a 
standard deviation of 0.056. It is important to indicate that 
average level of efficiency obtained here is slightly higher 
than that reported by previous studies. Bravo-Ureta et al. 
(2007) show, in their meta-regression analysis of TE in 
agriculture, an 84% average TE for stochastic frontier 
studies focusing on dairy farms in developed countries. 
However, higher levels of TE to those found in this study 
can be found in Alvarez and del Corral (2010), Abdulai 
and Tietje (2007) and Richards and Jeffrey (2000) for 
dairy farms in Spain, Germany, and Canada, 
respectively. 
 

 

Inefficiency analysis 

 

In the second step,  TI  is regressed  on  different farm‟s 

 
 

 

managerial characteristics using the quantile regression 
technique. Quantile regression models the relationship 
between inefficiency and the farm‟s characteristics using 
the conditional quantile, allowing the evaluation of the 
specific impact of these characteristics on different 
groups of farms clustered on their level of efficiency. The 
results of the quantile regression are presented in Table  
3. Given the inverse relationship between TI and TE 
(Equation 2), and because TI is the dependent variable in 
this analysis, a negative effect on TI has a positive impact 
on TE.  

The empirical results show that TMR estimate is 
positive but not statistically different from zero for the 
10th, 20th, and 30th quantiles. In other words, TMR does 
not significantly affect the efficiency of the most efficient 
dairy farms. In contrast, for less efficient ones (60th to 
90th TI quantiles), the parameter estimate is negative and 
statistically significant; which shows improvements in TE 
by adopting the TMR system. Similarly, the ratio of ‘feed 
per cow‟ has a positive and statistically significant impact 
on TE for the less efficient dairy farms and a positive but 
not significant impact for most efficient ones. These 
results imply that, in general, an increase in the 
intensification of a farm leads to improvements in TE, 
especially among less efficient farms, which is consistent 
with the literature (Cabrera et al., 2010; Alvarez et al., 
2008; Kompas and Chu, 2006).  

The use of bST for lactating cows has the effect to 
increase TE as revealed by the negative parameter 
estimate of this variable. This is not surprising since 
Bauman et al. (1999) show that the use of bST increases 
milk production and feed efficiency. It is worth noticing 
that this result does not depend on the type of the dairy 
farm as the parameter estimate is positive for all 
inefficiency quantiles.  

In contrast, as milking frequency increases, TE 
decreases for all quantiles as indicated by the positive 
and statistically significant parameter estimate of this



 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

 

F
 r

 e
 q

 u
 e

 n
 c

 y
 

 
 
 

 
120   
 
 
 

 
100  
 
 
 
 

 
80 

 
 
 

 
60 

 

 
40 

 
 
 

 
20 

 
 
 

 

0        
0.65 0.7 0.75 0.8 0.85 0.9 0.95 1  

Estimated Technical Efficiency 
 

Figure 1. Distribution of technical efficiency. 
 
 

 

variable. This result contradicts some previous studies 
Erdman and Varner (1994) who reported 3.5 to 4.9 
kg/day increase in milk production when cows are milked 
3 and 4 times daily compared to only 2 times milking in a 
day. However, Cabrera et al. (2010) argue that additional 
milk frequencies imply additional labor and additional 
feed intake which might affect the level of efficiency of the 
farm depending on the market conditions and farm 
characteristics.  

In relation to the milking system used in the farm, the 
results show that relative to pipeline parlor (the omitted 
variable), the use of flat barn and pit parlor increases 
dairy farms‟ inefficiency as indicated by positive 
parameter estimates of these variables. For instance, the 
effect of flat barn on inefficiency increases as we move 
from the most TE dairy farms to the less efficient ones. 
Table 3 shows that the negative effect of flat barn on 
technical efficiency is more than 345 fold for the 10th TE 
percentile than for the 80th TE percentile. Similarly, the 
effect of pit parlor is accentuated as dairy farms become 

 
 
 

 

less efficient. However, the parameter estimates for both 
milking systems are not statistically significant for the 
upper TE quantiles. In terms of the housing type, our 
results indicate that the type of housing has no significant 
impact on TE, which is consistent with Bewley et al. 
(2001), Hallan and Machado (1996), and Cabrera et al.  
(2010).  

One of the goals of this study was to assess the effect 
of the government payments on TE by type of dairy 
farms. Overall, government payments have a positive and 
statistically significant effect on TE for farms in the lower 
TE quantiles. However, these payments have no 
statistically significant effect on the TE of dairy farms that 

are already close to the frontier, the 90
th

 quantile. As we 

move far from the frontier, the effect of government 
payments on TE increases. In fact, this effect on TI is - 
0.007 for the 10th TI quantile, while it is -0.061 for the 
upper 90th TI quantile. In other words, the effect of 
government support on the TE of the less efficient farms 
is more than eight times higher than the effect on most 



            
 

 Table 3. Inefficiency analysis (N=1.151).          
 

            
 

 
Variables

1
 

    Quantiles      
 

 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9  
 

   
 

 Constant 0.0410*** 0.0496*** 0.0550*** 0.0680*** 0.0752*** 0.1063*** 0.1321*** 0.1824*** 0.2533***  
 

 TMR dummy 0.0048 0.0041 0.0033 -0.0029 -0.0055 -0.0088* -0.0059 -0.0160* -0.0380***  
 

 Waste handling 0.0035 0.0110*** 0.0118*** 0.0113*** 0.0178*** 0.0199*** 0.0251*** 0.0355*** 0.0408***  
 

 Flat barn -0.0018 0.0001 0.0038 0.0140 0.0173** 0.0256*** 0.0272*** 0.0358*** 0.0345***  
 

 Pit parlor -0.0046 0.0014 0.0095 0.0132* 0.0183*** 0.0140* 0.0194** 0.0328*** 0.0440***  
 

 Milking frequency 0.0047 0.0107*** 0.0159*** 0.0218*** 0.0249*** 0.0256*** 0.0267*** 0.0341*** 0.0384***  
 

 bST (%) -0.0003*** -0.0004*** -0.0004*** -0.0005*** -0.0005*** -0.0006*** -0.0007*** -0.0007*** -0.0006***  
 

 Feed/cow ratio 0.0000 0.0000*** 0.0000 0.0000 -0.0000 -0.0000*** -0.0000*** -0.0000*** -0.0001***  
 

 Free stall housing 0.0049 0.0075* 0.0076 0.0058 0.0055 0.0021 -0.0035 -0.0062 -0.0158*  
 

 Gov. payments -0.0075 -0.0091 -0.0082 -0.0027 -0.0002 -0.0128 -0.0180 -0.0320* -0.0605***  
 

 Nonfarm income 0.0018 0.0039 0.0168*** 0.0278*** 0.0374*** 0.0420*** 0.0486*** 0.0464*** 0.0496***  
 

 Calves sold 0.0204*** 0.0209*** 0.0234*** 0.0282*** 0.0301*** 0.0312*** 0.0338*** 0.0405*** 0.0468*  
 

 Crop sold 0.0201*** 0.0252*** 0.0313*** 0.0368*** 0.0435*** 0.0510*** 0.0485*** 0.0620*** 0.0630***  
 

 Family savings -0.0070 -0.0118*** -0.0192*** -0.0240*** -0.0257*** -0.0212*** -0.0215*** -0.0315*** -0.0340***  
 

 Investment/cow -0.0741*** -0.1162*** -0.1034*** -0.1492*** -0.1550*** -0.2285*** -0.2696*** -0.3726*** -0.4578***  
 

 Debt/cow 0.1532*** 0.1052*** 0.1999*** 0.2316*** 0.2818*** 0.3082*** 0.4050*** 0.2206* 0.2761*  
  

*P < 0.10; **P < 0.05; ***P < 0.01. 
1
 Dependent variable is the farm technical inefficiency, u. 

 
 

 

efficient farms. This outcome is very interesting for 
policy makers: less efficient dairy farms would 
benefit more from government payments than 
more efficient dairy farms.  

The results show that nonfarm income has a 
negative effect on farm efficiency regardless of the 
TI quantiles. This finding is consistent with the 
argument that off-farm work negatively affects 
agricultural production. Bravo-Ureta et al. (2006) 
explain that off-farm activities reduce the time 
available for agricultural work and that farmers 
involved in nonfarm activities are less concerned 
about improving the productivity and efficiency of 
their farm. This argument is confirmed by the 
positive and significant effect of family labor on the 
farm TE. Similar conclusions can be drawn for 

 
 
 

 

the income provided by activities other than dairy 
farming, such as calves and crops sales. These 
activities have a negative and statistically 
significant impact on TE. Moreover, the effect is 
more accentuated as we move from upper TE 
farms to lower ones.  

In addition, the financial health of the dairy 
farms plays an important role in TE. The results of 
this study indicate that as the investment per cow 
increases, TE also increases for all quantiles. 
Moreover, this increase is more accentuated for 
lower levels of technical efficiency. In contrast, as 
the debt per cow increases, TE decreases, 
especially for lower level quantiles. Finally, the 
level of family savings has also a positive effect on 
TE, with an accentuated effect for lower level 

 
 
 

 

quantiles. This is may be due to the fact that 
families with higher savings are able to invest 
more on their farms and contract less debt than 
the ones with lower savings. 
 

 

Conclusion 

 

This study analyses the determinants of TE 
among a sample of dairy farms in the State of 
Wisconsin. The results show that the deter-
minants of TE affect in very specific ways farmers 
with different levels of TE. This result confirms our 
hypothesis on the importance of controlling the 
farm heterogeneity when analyzing the deter-
minants of TE. The results of this study indicate 



 
 
 

 

that feeding factors, such as the use of TMR and feed per 
cow, affect positively the TE of dairy farms with lower TE 
levels, and negatively or they do not have an impact on 
TE efficiency of farms with higher TE levels. Another 
interesting finding is that, although all dairy farms would 
benefit from government payments, government 
payments contribute more to the increase of TE of dairy 
farms with lower TE levels than farms with higher TE 
levels. For example, the effect of the government 
payments on TE is about eight fold higher for lower TE 
farms than for the higher TE farms.  

The results also show that income either from non-farm 
activities or activities other than dairy farming have a 
negative effect on TE efficiency, regardless of the farm 
type. In addition, the farm‟s financial health plays an 
important role in technical efficiency. Technical efficiency 
increases as family savings and investment per cow 
increase and decreases as debt per cow increases. 
 

 

ACKNOWLEDGEMENT 

 

The authors are thankful to the University of Wisconsin 
Center for Dairy Profitability (http://cdp.wisc.edu/) for 
providing the data used to perform the present study. 
 

 
REFERENCES 
 
Abdulai A, Tietje H (2007). Estimating technical efficiency under 

unobserved heterogeneity with stochastic frontier models: application 
to Northern German dairy farms. Eur. Rev. Agric. Econ., 34: 393-416.  

Alvarez A, del Corral J, Solís D, Pérez JA (2008). “Does Intensification 
Improve the Economic Efficiency of Dairy Farms?”. J. Dairy Sci., 
91(9): 3699-3709  

Ball E (2009). Agricultural productivity in the United States. USDA ERS. 
http://www.ers.usda.gov/Data/AgProductivity/#statetables. Accessed 
November 2010.  

Bauman DE, Everett RW, Weiland WH, Collier RJ (1999). Production 
responses to bovine somatotropin in northeast dairy herds. J. Dairy 
Sci., 82: 2564-2573. 

 
 
 
 

 
Bravo-Ureta B (1986). Technical efficiency measures for dairy farms 

based on a probabilistic frontier function model. Canadian J. Agric. 
Econ., 34: 399-415.  

Bravo-Ureta B, Solís D, Moreira V, Maripani J, Thiam A, Rivas T (2007). 
Technical efficiency in farming: a meta-regression analysis. J. Prod. 
Anal., 27: 57-72.  

Bravo-Ureta B, Solís D, Cocchi H, Quiroga R (2006). The impact of soil 
conservation and output diversification on farm income in Central 
American hillside farming. Agric. Econ., 35: 267-276.  

Bewley J, Plamer RW, Jackson-Smith DB (2001). Modeling milking 
production and labor efficiency in modernized Wisconsin dairy herds. 
J. Dairy Sci., 84: 705-716.  

Cabrera  V,  Solís  D,  Del-Corral  J  (2010).  Determinants  of  technical 
efficiency among dairy farms in Wisconsin. J. Dairy Sci., 93: 387-393. 
Cabrera  V,  Hagevoort  R,  Solís  D,  Kirksey  R,  Diemer  J  (2008).  

Economic impact of milk production in the State of New Mexico. J. 
Dairy Sci., 91: 2144-2150.  

Caudill SB, Ford JM, Gropper DM (1995). Frontier estimation and firm-
specific inefficiency measures in the presence of heteroscedasticity. 
J. Bus. Econ. Stat., 13: 105–111.  

Chen C (2005). Growth charts of body mass index (BMI) with quantile 
regression. Proceedings of International Conference on Algorithmic 
Mathematics and Computer Science, Las Vegas.  

Demir N, Mahmud S (2002). Agro-climatic conditions and regional 
technical inefficiencies in agriculture. Can. J. Agric. Econ., 50: 269-
280. 

Erdman RA, Varner M (1994). Fixed yield responses to increased 
milking frequency. J. Dairy Sci., 78: 1199-1203. 

Hallan D, Machado F (1996). Efficiency analysis  with panel data: A 
study of Portuguese dairy farms. Eur. Rev. Agric. Econ., 23: 79-93.  

Koenker R, Hallock KF (2001). Quantile regression. J. Econ. 
Perspectives, 15: 143-156.  

Kompas T, Chu T (2006). Technology choice and efficiency on 
Australian dairy farms. Aust. J. Agric. Resour. Econ., 50: 65–83.  

Murova O, Chidmi B (2009). Impacts of Federal Government programs 
and specific farm variables on technical efficiency of dairy farms. 
Paper presented at the 2009 Southern Agricultural Economics 
Association Annual Meeting, Atlanta, Georgia (available at 
http://purl.umn.edu/46822).  

Newman C, Matthews A (2006). The productivity performance of Irish 
Dairy farms 1984-2000: a multiple output distance function approach. 
J. Prod. Anal., 26: 191-205.  

Tauer LW (2001). Efficiency and competitiveness of the small New York 
dairy farm. J. Dairy Sci., 84: 2573-2576.  

Tauer LW, Mishra AK (2006). Dairy farm cost efficiency. J. Dairy Sci., 
89: 4937-4943. 


