.3
.4)..-

”a

I:

I . ll?‘ .
.z .
. . . . .0 “0,09. .
. , “that. Human til
, a
flh.nc...sa.n~% ..
.‘Vumwéflh
....:h..§.,bn a
. . LN. Ema.

1r. "3.... . 1
35¢
. .a.
.155...» mflnmyflfl
. .ru.V :3 :1.
‘ 2‘12!
.Hfifluu.»

5'9...

. ’14:! L
T x xxiva WM“? .53“.
£...~.?£.«. fag.” .z
. iii? 5::

V Sh

“'13"

5311» I

)

1 .l.

...—8

at} .
1 .i

L I .\ 4 z.
”3.... .7
21!th
i [.5313- 3
1. :3)... A.)

 

 

 

. V 1
.1. . 1+1...“ I! ‘ ... .3
51:2:
.1... .. 1: 2.31....
:0 . i... (1.1;:
y {it}

 

3.. E!
i-ln. it
£ ....~v..!

4).. 0.0-5.

unulo'ii .
§§%.. .
;, 4....5

 

y 2... 4 .
3:13,. ..
a. .hfimacflzr

Ii!

uthultc
an.

+1
cu

“.4
jun

i.

 

/ LIBRARY
fill! Michigan State
University

This is to certify that the
thesis entitled

ECONOMIC ANALYSES OF REPRODUCTION
MANAGEMENT STRATEGIES AND TECHNOLOGIES ON
U.S. DAIRY FARMS

presented by

NICOLE J. OLYNK

has been accepted towards fulfillment
of the requirements for the

MS. degree in Agricultural, Food, and
Resource Economics

 

 

flW

L/ Major Professor’ Sig ture
4M3» F (0% ZOcflX

Date

MSU is an afiinnative-action, equal-Opportunity employer

a.-----------.---—-.—-.-----—

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

5/08 KlProj/AccauPrelelRC/DateDuemdd

ECONOMIC ANALYSES OF REPRODUCTION MANAGEMENT STRATEGIES
AND TECHNOLOGIES ON U.S. DAIRY FARMS
By

Nicole J. Olynk

A THESIS
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
MASTER OF SCIENCE
Department of Agricultural, Food, and Resource Economics

2008

ABSTRACT
ECONOMIC ANALYSES OF REPRODUCTION MANAGEMENT STRATEGIES
AND TECHNOLOGIES ON U.S. DAIRY FARMS
By

Nicole J. Olynk

Reproductive management of dairy cattle is crucial to whole-farm profitability as
it enables milk sales, provides replacement animals, and is a factor in culling decisions.
The dairy industry has responded to challenges in managing dairy cattle reproduction
with innovative technologies and reproductive management programs that enable
producers to synchronize ovulation, thereby lessening or eliminating the need for visual
heat detection, or to make heat detection more efficient through the use of aids or
automated computer-based record keeping systems. Dairy producers today face
decisions regarding which reproductive management program is optimal for their farm
operation. The analyses presented built upon prior reproductive management studies and
sought to inform economically sound decision making regarding reproductive program
and technology adoption. The varying costs and revenues associated with reproductive
performance across farms illustrated the need for farm-specific analysis regarding
selection of economically optimal reproductive management programs. Through the use
of surveys, sensitivity analyses to reproductive program costs, and assessment of farm
manager decision making under risk, the reproductive program decisions made on farms

are better understood.

ACKNOWLEDGEMENTS

I would like to thank my committee, namely Dr. Christopher Wolf, Dr. Stephen
Harsh, and Dr. Richard Pursley for their guidance in this research. Special thanks go to
Dr. Wolf, my research advisor and major professor. His encouragement, advice, and
friendship have been invaluable in my graduate career thus far.

I would like to thank all of the graduate students and faculty with whom I have
worked with at Michigan State, and who have been so helpful and supportive.
Specifically, I thank Nicky Mason on whom I can always rely, regardless Of the task or
time of day. Lee Schulz deserves special thanks for his help and continuous support with
this work, specifically with Chapter 4. Lee’s patience and time in helping with Chapter
4, specifically, are greatly appreciated. I thank J oleen Hadrich for her willingness to help
and availability to bounce ideas off of. More generally, I would like to thank all of my
friends for all of their support and encouragement.

Finally, I would like to thank my family for their unending support. A special
thanks goes to my fiancé, Dave, for his constant companionship, support, and ability to
put things into perspective. I am forever grateful for his support and willingness to help

in any way he can; his support has been invaluable.

iii

TABLE OF CONTENTS

LIST OF TABLES .............................................................................................................. v
LIST OF FIGURES ........................................................................................................... vi
CHAPTER 1: GENERAL INTRODUCTION ................................................................... 1
CHAPTER 2: ECONOMIC ANALYSIS OF REPRODUCTIVE MANAGEMENT
STRATEGIES ON U.S. COMMERICAL DAIRY FARMS ............................................. 5
2.1 Introduction ............................................................................................................... 5
2.2 Materials and methods .............................................................................................. 7
2.3 Results and discussion ............................................................................................ 14
2.4 Conclusion .............................................................................................................. 34

CHAPTER 3: DECISION SUPPORT FOR REPRODUCTIVE MANAGEMENT
TECHNOLOGY AND PROGRAM ADOPTION ON U.S. COMMERICAL DAIRY

FARMS ............................................................................................................................. 36
3.1 Introduction ............................................................................................................. 36
3.2 Materials and methods ............................................................................................ 40
3.3 Results and discussion ............................................................................................ 51
3.4 Conclusion .............................................................................................................. 58

CHAPTER 4: A STOCHASTIC ECONOMIC ANALYSIS OF DAIRY CATTLE
ARTIFICAL INSEMINATION REPRODUCTIVE MANAGEMENT PROGRAMS 60

4.1 Introduction ............................................................................................................. 60
4.2 Materials and methods ............................................................................................ 65
4.3 Results and discussion ............................................................................................ 76
4.4 Conclusion .............................................................................................................. 82
CHAPTER 5: SUMMARY AND DISCUSSION ............................................................ 84
APPENDICES .................................................................................................................. 88
APPENDIX 1: DAIRY PRODUCER SURVEY .......................................................... 89
APPENDIX 2: SYNCHRONIZATION PROGRAM REFERENCE SHEET
INCLUDED WITH SURVEY .................................................................................... 101
REFERENCES ............................................................................................................... 104

iv

LIST OF TABLES

Table 2 a. Summary of survey responses regarding facilities used to house cows and
heifers ........................................................................................................................ 16

Table 2 b. Summary of survey responses to questions regarding paid and unpaid labor
usage in 2005 with corresponding means (number of responses = 35) .................... 17

Table 2 c. Summary of survey responses to questions regarding general reproductive
management parameters with corresponding means ................................................ 19

Table 2 (1. Summary of survey results to questions regarding heat detection methods used
for cows and heifers (number of responses = 82 and 52 for cows and heifers,
respectively) .............................................................................................................. 22

Table 2 e. Summary of survey results regarding reasons why synchronization programs
were not used for cows or heifers (number of responses = 25 and 43 for cows and

heifers, respectively) ................................................................................................. 25

Table 2 f. Summary of survey responses regarding synchronization programs used
(number of responses = 53 and 15 for cows and heifers, respectively) .................... 26

Table 3 a. Calculating cumulative probability of pregnancy (AI submission rate = 50%,

CR= 38%) ................................................................................................................. 42
Table 3 b. Calculating cumulative probability of pregnancy (AI submission rate = 100%,
CR= 35%) ................................................................................................................. 42
Table 3 c. Program cost sensitivity to cutoff rule in example scenario ........................... 54

Table 3 d. Reproductive management program values for first lactation cows (26 day

period for Ovsynch) .................................................................................................. 55
Table 3 e. Reproductive management program values for first lactation cows (33 day
period for Ovsynch) .................................................................................................. 56
Table 3 f. Reproductive management program values for third lactation cows (26 day
period for Ovsynch) .................................................................................................. 57
Table 3 g. Reproductive management program values for third lactation cows (33 day
period for Ovsynch) .................................................................................................. 58
Table 4 a. On-F arm values affecting costs by farm size .................................................. 75
Table 4 b. Summary statistics for program costs in example analysis ............................. 77

LIST OF FIGURES

Figure 2 a. Artificial insemination program costs: sensitivity to heat detection rates and
heat detection efficiency ........................................................................................... 33

Figure 2 b. Artificial insemination program costs: sensitivity of Ovsynch synchronization

program to time per injection .................................................................................... 34
Figure 4 a. Ovulation synchronization programs used as examples ................................ 73
Figure 4 b. Mean-variance analysis for example .............................................................. 78
Figure 4 c. Cumulative density functions for example analysis ....................................... 79
Figure 4 d. Small farm sensitivity analysis ...................................................................... 80
Figure 4 e. Sensitivity analysis: Ovsynch incurring additional breeding cost ................. 81

vi

CHAPTER 1: GENERAL INTRODUCTION

Dairy cattle reproductive efficiency is closely tied to the profitability of
commercial dairy operations in the United States. Dairy farm profitability is affected
through various factors which are dependent on reproductive performance, including
milk production, number of replacements, voluntary and involuntary culling, breeding
costs, and costs associated with veterinary care (Britt, 1985). Given the integral link
between dairy cattle reproductive efficiency and total farm profitability, dairy producers
have sought technologies and programs which facilitate efficiently managing cattle
reproductive performance.

Recent trends towards decreased reproductive performance industry-wide have
led to increased focus on development of reproductive management programs and
technologies. Specifically, increased herd sizes and milk production levels have affected
how dairy farms are managing dairy cattle reproduction (Pursley et al., 1997). Today,
several reproductive technologies are available for use on commercial dairy farms,
including artificial insemination (AI), estrus and ovulation synchronization programs,
sex-sorted semen, pedometers, computer-based record systems, and multiple visual and
electronic estrus detection aides. F arm-specific factors, including varying on-farm input
costs, facilities used to handle and house cattle, previous levels of reproductive
performance achieved, management ability, and knowledge cause costs and reproductive
performance outcomes to vary on a given reproductive management program.

The operator’s degree of risk aversion, financial positioning of the farm,

availability of or access to information on available technologies, and the risk levels

associated with the outcomes of the technology are a few of the factors that affect
adoption of technologies on dairy farms. Additionally, labor availability, labor costs,
ease of cattle handling, and previous or baseline measures of reproductive performance,
may be influential in determining the profit maximizing reproductive management
program for a farm operation. Certainly the program that is found to be optimal for a
farm with one set of characteristics may not be optimal for a farm with a different set of
characteristics.

Complicating the technology and program adoption decisions of dairy farmers is
the fact that costs associated with and results expected from reproductive management
technologies and programs are uncertain in many cases, and are variable across
individual farms. Uncertainty about the performance of new technologies arises not only
from a lack of performance history, but also from a lack of knowledge, which may be
caused by asymmetric information. Dairy producers facing uncertainty in reproductive
program outcomes could certainly benefit fi'om decision-support tools which allow
sensitivity analysis to key outcome parameters. With a user-friendly tool available to
perform sensitivity analysis for programs or technologies for which the dairy might
consider adoption, the dairy farm managers would be able to determine the range of
outcomes that may be expected. The ease of performing such sensitivity analysis allows
producers to make more informed adoption decisions when considering reproductive
management programs.

This series of analyses begins by seeking to aid in understanding dairy farm
decision support needs regarding decision making on reproductive technology and

program adoption through surveying of dairy farmers across multiple states. Then, armed

with the information recovered through survey analysis, a user-fn'endly decision support
tool, designed for on-farm use, was developed to address the needs of dairy farmers as
they make decisions regarding reproductive management. Finally, to address the
heterogeneous risk preferences among dairy farmers, efficient sets of reproductive
management programs were identified for producers within broad general categories of
risk preference. Given the analysis presented consists of multiple-steps, a series of
objectives are highlighted for each portion of the analysis.

Overall, the objectives of this series of analyses include identification of key
issues for dairy farmers through surveying dairy managers, the development of a user-
friendly decision-support tool, and finally the assessment of efficient sets of reproductive
management programs for farms with various characteristics. This analysis uses survey
data from US commercial dairy operations to provide economic insight into reproductive
management program and technology adoption decisions. Specifically, survey data was
collected and used to inform the economic analysis of various reproductive management
program decisions, to aid in identifying factors affecting whether farms used reproductive
management programs, and if so, to determine which programs a farm with given
characteristics was likely to choose. After highlighting key farm characteristics that
affect reproductive technology and management program decisions, a tool was developed
that allows farm-Specific parameters to be entered and used in evaluating reproductive
management decisions. Additionally, since the decision tool allows farm managers to
enter cow numbers per group rather than assessing all cows on the farm at once, farm
managers can determine the optimal program for different groups of cows on their

operation. To determine the economically optimal programs for dairy farms with various

characteristics (i.e., farm size, risk preferences of farm managers, on-farm reproductive
program costs) stochastic dominance was employed. Due to the heterogeneity of risk
preferences among dairy farmers, stochastic dominance was utilized in order to separate
the sets of risky options to identify the efficient sets of reproductive management
programs for decisions makers with specific risk preferences.

As dairy farm profitability continues to rely on reproductive performance and
efficiency, and increased production levels coupled with increasing farm sizes lead to
challenges in managing reproductive performance, dairy managers can benefit from
decision-support in identifying the economically optimal reproductive programs and
technologies available.

This thesis proceeds as follows: an economic analysis, parameterized using
survey data, highlights the reasons why farms with different characteristics select various
reproductive management programs and technologies in chapter 11. Survey analysis and
assessments of the sensitivities of different types of reproductive management programs
to varying on-farm costs and labor efficiencies are used to highlight why farms select the
reproductive management programs that they do. Chapter III depicts a user-friendly
spreadsheet tool which was developed for on-farm decision support regarding selection
of reproductive management programs. Chapter IV describes the efficient sets of
reproductive programs for dairy farms with varying characteristics under first and second

degree stochastic dominance.

CHAPTER 2: ECONOMIC ANALYSIS OF REPRODUCTIVE MANAGEMENT
STRATEGIES ON U.S. COMMERICAL DAIRY FARMS
2.1 Introduction
Reproductive performance on the dairy farm affects the farm profitability through
milk production, number of replacements, voluntary and involuntary culling, breeding
costs, and costs associated with veterinary care (Britt, 1985). The economic implications
of reproductive management decisions are critical given the link between dairy herd
management and reproductive performance. Today, many reproductive technologies are
available for use on commercial dairy farms with the costs and reproductive performance
associated with these technologies varying considerably across farms. Farm reproductive
management programs differ due to varying on-farm costs, facilities, farm goals and
values, and management styles. These factors, in addition to labor availability, cost of
labor, ease of cattle handling, and previous levels of reproductive performance, determine
profit maximizing reproductive management techniques and technologies for dairy herds.
Several survey-based studies in recent years focused on dairy herd reproductive
performance and management practices. providing a great deal of information about the
current practices, performance, and management techniques of dairy farms. These
overviews are useful for dairy producers, extension educators, researchers, and related
farm service industries as they provide current information regarding what practices are
actually being adopted and used on commercial dairy operations. However, additional
analysis is necessary to understand farm decisions relative to reproductive management

programs and the resulting economics.

A recent survey across multiple states by Caraviello et al. (2006) analyzed 153
large US dairy herds in the Alta Genetics Advantage Progeny Testing Program in 2004.
Caraviello et al. (2006) asked questions regarding general management, sire selection,
reproductive management, inseminator training and technique, heat abatement, body
condition scoring, facility design and grouping, nutrition, employee training and
management, and animal health and biosecurity. Of the 103 herds which completed the
survey, the average herd size was 613 cows, and 87% of those herds utilized hormonal
synchronization or timed artificial insemination (TAI) in their reproductive management
programs. Caraviello et al. (2006) provided an in-depth reference of management
practices being used on large commercial US dairy herds in 2004 and a valuable resource
for benchmarking or comparison purposes.

Meadows et al. (2005) found through the use of a Spreadsheet-based model that
“. . .inefficient reproduction becomes marginally more costly to producers as performance
declines and warrants increased attention.” Meadows et al. (2005) also found that there
existed decreasing marginal benefits to improved reproduction as reproductive
performance improves. These decreasing marginal benefits to reproductive performance
improvements may explain why different farms use different reproductive management
strategies. Those farms currently achieving high levels of reproductive performance may
have less incentive to initiate a potentially performance enhancing change than a farm
with sub-par current performance.

A survey of bovine practitioners was conducted to evaluate the cost effectiveness
of systematic breeding programs (Nebel and Jobst, 1998). Using the values found

through their survey Nebel and J obst (1998) calculated estimated costs per pregnancy for

Ovsynch and Targeted BreedingTM (Pharmacia-Upjohn, Kalamazoo, MI). Further, Nebel
and Jobst (1998) conclude that systematic breeding program decisions must be evaluated
for cost effectiveness in order to determine the optimal program.

The objectives of this study were to utilize survey data to provide economic
insight into why varying types of farms used different reproductive management
programs, and to identify those factors affecting whether farms use various reproductive
technologies. This analysis sought to build upon prior reproductive management studies
and dairy industry surveys by using survey data to inform the economic analysis of
various reproductive management programs. Survey data was used to parameterize the
economic analysis and inform the discussion regarding economic and management

implications of reproductive management decisions.

2.2 Materials and methods

A survey was mailed to 1,000 dairy farms in Michigan, New York, Texas,
Wisconsin, and Florida between August and December of 2006. The survey was
developed to obtain data regarding reproductive management and performance in 2005
and is displayed in Appendix 1. This analysis ultimately sought to identify the factors
affecting farm reproductive management program adoption decisions and to explore the
management and economic implications behind various reproductive management
programs.

Dairy farms receiving surveys were selected randomly from those permitted to
sell milk in the aforementioned states, thereby allowing a broad range of farms to

participate in the survey. Out of the 1,000 surveys mailed, the number of farms receiving

surveys in each state was selected proportionately to the total number of dairy farms in
that state. A total of 102 surveys were returned, resulting in a 10.2% response rate. Only
those respondents who were actively operating dairy farms in 2005 and chose to
participate in the survey were included in this analysis, resulting in a total of 60 potential
respondents for each question. Consistent with Michigan State University research
requirements when administering a survey, respondents were presented the option to
decline to answer individual questions or sections of the survey at their discretion, if they
chose to participate at all.

The random selection of farms that received the survey allowed equal opportunity
for selection regardless of participation in various farm programs or membership in a
particular cooperative. The negative outcome of using such a selection process, where
farms are drawn randomly from a diverse population, and likely with a perceived low
incentive to participate, was a lower response rate. Although farms were randomly
selected to receive surveys, given the relatively small sample size and response bias
inevitably introduced with mailed surveys, the sample was not expected to be
representative of the diverse population of US dairy farms. However, the survey data
itself was not the primary focus of this analysis. The survey data collected was used to
parameterize the analysis of factors affecting the decisions of farms to use various
reproductive management programs. In addition, management and economic
implications of various reproductive management programs were explored.

The survey included questions about dairy reproductive management and
performance of both heifers and cows on the operation in 2005. Questions relating to

general farm and operator characteristics, including cow numbers, record keeping

methods, labor costs, and culling were asked in order to better understand the
characteristics of the farms which used various reproductive management techniques.
More in-depth questions were then asked in sections surrounding reproductive
management and performance, heat detection methods, synchronization programs, and
recent reproductive management changes implemented on the farm. A description of
Ovsynch, Presynch with Ovsynch, Heatsynch, Cosynch, controlled internal drug-
releasing intravaginal insert (CIDR) containing progesterone with PGFZa, and the
Targeted Breeding Protocol were provided as an appendix to the reproductive
management survey for reference and is included in Appendix 2.

Summary statistics were computed for continuous variables. Throughout the
results, the “number of total responses” accompanies summary statistics, which indicates
the total number of usable responses to a given question. Many questions allowed a
respondent to check all answers which were applicable to the operation from a multiple
choice list, and such questions were analyzed by tabulating the total number of responses
and computing frequencies.

The survey described above was used in order to inform the economic based
assessment of the use of various reproductive management techniques and technologies.
Economic assessments were based upon the underlying assumption that respondents were
seeking to maximize their individual profitability through their management decisions.

Budgets were developed in Excel (Microsoft, Seattle, WA) with the
understanding that the costs associated with achieving various levels of reproductive
performance and with the administration of reproductive management programs will vary

across farms. For example, synchronization program costs include the cost of hormones,

supplies, and the labor needed to administer the injections. Time required to administer
injections is a function of facilities employed and the skill level of the person
administering the treatments. Additionally, visual heat detection program costs vary
depending on the hourly labor costs for those people performing the visual heat detection
and the efficiency with which they detect heats.

Reproductive program costs were calculated on a per cow basis to facilitate
comparison across different program types. Synchronization program costs were
calculated on a per cow basis, although visual heat detection costs had to be adjusted to
obtain a per cow program cost. Heat detection program costs were adjusted by dividing
the cost of heat detection for a group of cows over the number of cows in the group. The
time value of money was ignored due to the relatively small time frame analyzed and
because programs were compared beginning at the same point in time. Therefore, a
program’s cost would only differ in timing due to increased number of services to
achieve a 90% cumulative probability of pregnancy. The resulting differences due to
time value of money were negligible for the analysis completed, which sought to
highlight relative sensitivity to specific on-farm costs among programs.

In calculating program costs, the total cost of each program was calculated by first
determining the number of months required to achieve a 90% cumulative probability of
pregnancy under that program’s resulting conception rate (CR) and heat detection rate
(HDR). The number of months necessary to achieve the 90% cumulative probability of
pregnancy will clearly depend on the pregnancy rate (PR) achieved monthly, which will
in turn be dependent on the CR and HDR achieved on-farm. By calculating the number

of months over which the program would necessarily be administered to achieve the

10

target cumulative PR, the total program cost can be calculated by multiplying the number
of months the program will be administered by the monthly cost of the program. The
services per conception were calculated for each program to achieve the cumulative
probability of pregnancy of 90%. This means that a cow was assumed to be bred
multiple times until an expected cumulative 90% chance of pregnancy occurred. Total
costs across programs were compared by calculating all program costs subject to
achieving the 90% cumulative probability of pregnancy.

The budgets allowed for entry of CR, HDR, labor efficiency in detecting heats or
giving injections, artificial insemination (AI) costs, synchronization program costs, and
the cost of labor. Additionally, the budgets allowed calculation of breakeven costs,
indicating at what labor cost farms with a given set of characteristics should switch fi'om
one program to another. Using the cost of labor as an example, holding other on-farm
costs constant, the authors calculated the labor cost below which the farm should utilize
visual heat detection and above which the farm should switch to a synchronization
program rather than use labor for heat detection tasks.

Heat detection program costs, assessed on a monthly per cow basis were
calculated as follows:

Heat Detection Program Cost/Cow/Month = [((TIME * OBS * 30.4) *
LABOR(HD))/COWS] + AID + (C1 * HDR),

where:
TIME = Minutes per day invested in heat detection for a single group of cows,

OBS = Number of times the group is observed per day,

LABOR<HD) = Cost of labor (in dollars per minute) to performs heat detection,

11

COWS = Number of cows in the group,
AID = Cost per cow per period of heat detection aid utilized,
CI = Cost of an artificial insemination, and

HDR = Heat detection rate.

Synchronization program costs, assessed on a monthly per cow basis were calculated as

follows:

Synchronization Program Cost/Cow/Month = (PGnRH "' XGnRH)+(PPGF2a * XPGFZG) +

(MIN *

where:

INJ * Labor (Inject)) + (C1 * HDR),

PGnRH = Cost of GnRH per injection,
XGnRH = Number of GnRH injections administered,
PpGFz‘Jl = Cost of PGFZa per injection,

XpGF2a = Number of PGF2a injections administered,

MIN = Minutes to give a single injection,

IN] = Total Number of injections in the series,

LABOR (Inject) = Cost of labor (in dollars per minute) to gives injections,
CI = Cost of an artificial insemination, and

HDR = Heat detection rate.

It should be noted that in the calculation of the program costs for both

synchronization and visual heat detection programs, the cost of labor to perform the

program was assessed according to the wage paid if the task was performed by paid labor

12

or through the opportunity cost of the labor if unpaid labor was used. The cost of paid
labor is easily accounted for via wage rates paid, although in order to assess the true costs
associated with various reproductive management programs a charge for the opportunity
cost of unpaid labor must also be included.

Using the heat detection and synchronization costs per cow per month, the total
breeding program costs were calculated by summing the program costs per month over
the number of months necessary to obtain a 90% cumulative probability of pregnancy.
For example, if based on the inputted CR and HDR it is calculated to take four months in
order to achieve a 90% cumulative probability of pregnancy, the total program cost was
the monthly program cost multiplied by four.

Sensitivity analyses were performed, holding all other variables constant, for
differing costs of labor, minutes required per injection, and efficiency of heat detection.
Heat detection labor efficiency was altered by holding the minutes of heat detection per
day constant while changing the resulting HDR. Efficiency in giving injections was
altered by adjusting the labor minutes required per single injection. Additionally,
breakeven labor costs were calculated to determine at which labor cost a synchronization
program becomes less costly than a visual heat detection program, holding all other
variables constant.

Baseline assumptions for CR, HDR, costs per AI, time spent on heat detection per
day, time required to give a single injection, and costs per injection for GnRH and PGFZa
were obtained from the survey averages. The baseline values used were a CR of 38%
(calculated using average services per conception (SPC) reported in survey of 2.66, as

l/SPC), HDR of 52% for visual heat detection, HDR of 100% under synchronization

13

protocols (because cows are assumed bred by TAI), cost per A1 of $17.30, time spent on
heat detection of 2.15 labor hours per day (calculated from 3 observations, on average, of
43 minutes per observation, on average), 2.1 minutes required to give a single injection,
and costs per injection of GnRH and PGFZa of $3.59 and $2.52 respectively. Program
cost comparisons and sensitivity analysis focus on Al strategies. The synchronization
protocol chosen for use in this example was Ovsynch, as it was the most common
program used among those using synchronization in the survey, with 38% of those
respondents using synchronization having employed Ovsynch for their cows. Ovsynch
consists of an injection of GnRH, followed 7 days later by PGFZa, followed 48 hours later
by a second GnRH injection, and followed by breeding 24 hours after the second GnRH
injection (Pursley et al., 1995). These baseline assumptions were used to create a
scenario in which sensitivity analysis could be performed to examine differences in costs

due to changes in the parameters used.

2.3 Results and discussion

Survey results were organized into sections regarding farm and operator
characteristics, reproductive management and performance, heat detection methods,
synchronization programs, and recent reproductive management changes. These survey
findings are discussed with regards to how they inform the underlying economic analysis
of factors affecting which farms use various reproductive management programs.
Economic and management implications were then parameterized using the survey
results. Economic analyses, including sensitivity analyses of program costs to labor

costs, visual heat detection efficiency, and time required to give injections, are presented

14

to highlight those factors affecting on-farm decisions regarding reproductive program
use.

The mean total herd size (including milking and dry cows) of survey respondents
was 238 cows with a range of 20 to 1588 cows (57 respondents). Herds of less than 200
cows accounted for 69% of the respondents. In 2005, 40.1% of operations in the US
dairy industry had less than 200 milk cows (NASS, 2007). Individual states varied
considerably in farm size distributions, with New York, Michigan, and Wisconsin having
had 53.5%, 45.5%, and 68.5% of their farms under 200 cows, respectively (NASS, 2007).

Number of heifers ranged from 0 to 1401 (56 respondents), with 7 farms reporting
the use of a custom heifer raiser. The average number of bulls on the farms was 2, with a
range of 0 to 13, and a mode of zero (53 respondents). The number of females per bull
varied greatly across farms, from 13.2 to 250 females per bull, which illustrated the
different ways bulls were used on these operations, ranging from solely natural breeding
to a clean-up bull for only limited numbers of animals or problem breeders. A commonly
used ratio is one bull for every 25 females (Fricke and Niles, 2003) and the range of
survey responses for the number of females per bull indicated a wide range of
reproductive management practices using natural service.

While a question specifically regarding whether the farm was an organic dairy
was not included in this survey, reasons provided in response to various management
related questions indicated whether the farm was managed organically. Of the 60
responses used in this analysis, 10% of respondents noted explicitly in survey responses

that they operated organic dairies.

15

Facilities affect reproductive management decisions regarding ease and efficiency
of heat detection and sorting cattle for administering shots or performing AI. Facilities
used for housing heifers and cows were reported separately by respondents through an
inclusive list which allowed respondents to select more than one housing option if
different housing types were used throughout the year. For example, those farms which
utilize stanchion barns for a portion of the year but also utilize pasture for a portion of the
year for their heifers would have two responses to this question for their heifers. Survey

results regarding housing facilities are provided in Table 2a.

Table 2 a. Summary of survey responses regarding facilities used to house cows and
heifers

 

 

 

 

 

 

 

 

 

 

 

Housing Type Cows Heifers
Percent of responses

Stanchion Barn 27 5
Freestall Barn 29 23
Bedded Pack Barn 7 24
Drylot 13 13
Pasture 24 34
Other 0 1
Total Number of Responses 105 107

 

 

 

 

In order to complete an economic assessment of the costs associated with various
reproductive management strategies, an estimate of the associated labor costs must be
calculated. Respondents were asked questions regarding numbers of paid and unpaid
employees and the salaries received by workers of the following levels: hired managers,
full-time workers, part-time workers employed throughout the year, seasonal workers,

and others not fitting any of these categories. Results regarding paid and unpaid labor

16

usage in 2005 are provided in Table 2b. Note that the data collected regarding family and
hired labor usage in this survey differs in structure from other surveys seeking data on
number of employees. For example, Caraviello et al. (2006) reported the number of
people involved in the management and operation of the farm, thereby providing valuable
insight into the cows per worker and other efficiency measures. For economic based
decision making, however, it is the total cost of all labor used that is important. The cost
of paid labor is easily assessed via survey responses for wage rates paid, although in
order to assess the true costs associated with various reproductive management programs

a charge for the opportunity cost of unpaid labor must also be included.

Table 2 b. Summary of survey responses to questions regarding paid and unpaid
labor usage in 2005 with corresponding means (number of responses = 35)

 

 

 

 

 

 

 

 

 

 

 

 

 

Worker Type Number Months Hours Wage rates
of worked worked per reported for paid
workers per year month labor
Hourly Annual
Pay Salary
Unpaid Labor
Spouse(s) of 1.3 11.6 235
Operator(s)
Children over 12 1.6 9.9 106
Other unpaid labor 1.8 7.4 104
Paid Labor
Hired 1.6 12 273 $12.78 $30,742
managers/operators
Full-time 4.75 1 1.9 220 $8.83 $23,360
Part-time 1.5 10.9 154 $8.16 $6,667
Seasonal 3.4 4.9 150 $8.71

 

 

 

 

 

 

 

Record keeping is integrally important to the success of a reproductive program.
Records regarding observed heats, treatments administered, behavior observed, or past

challenges aid in efficient decision making on the dairy operation. Respondents were

17

asked about all of the herd management record keeping systems used on the farm. Sixty-
nine responses from 60 farms were received regarding record keeping, indicating that
some farms utilized more than a single method. Overall 41% of responses were for paper
records, 28% used their Dairy Herd Improvement Association, 14% used PC Dart (Dairy
Records Management Systems, www.drms.org/pcdarthtm), 12% used Dairy Comp 305
(Valley Agriculture Software, Inc., Tulare, CA) or Scout (Valley Agriculture Software,
Inc., Tulare, CA), and 6% used another method. The most common entry for a method
other than those options given was an in-house developed Excel (Microsoft, Seattle, WA)
spreadsheet designed for individual herd record keeping.

Survey questions regarding cull rates and reasons for culling were asked,
primarily to assess how much culling was due to reproductive failure. The average cull
rate reported was 20.5% across the 55 herds which responded to this question, with
reported figures ranging from 0 to 41%. The average cull rate of 20.5% was lower than
expected, although previous studies have found that average cull rates increase with herd
size (Hadley et al., 2006), and 69% of the sample is comprised of herds with less than
200 cows. On average, of the 31 respondents answering this question, 19% of total culls
were due to poor reproductive performance. This finding is similar to the previous
findings of Hadley et al. (2006) who indicated that 18.9% of total culls were attributed to
reproductive performance across all of the states included in their study. Clearly, with
nearly a fifth of culls being attributed to reproduction, the reproductive performance of
the dairy farm has far-reaching implications for not only reproductive efficiency, but

culling patterns as well.

18

Survey data collected on overall reproductive management and performance
included average age of heifers at first service and first calving, days open, calving
interval, voluntary waiting period, SPC for heifers and cows, and length of dry period.
The average calving interval found in this survey was 13.1 months, which is similar to
the previously published survey data of Caraviello et al. (2006), who reported 13.8
months as the average calving interval. The average SPC reported in this survey were
2.66 for cows and 1.8 for heifers. A summary of the responses to the questions regarding
general reproductive management, including the number of respondents answering each

individual question, is provided in Table 20.

Table 2 c. Summary of survey responses to questions regarding general
reproductive management parameters with corresponding means

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Question Response Number of
respondents
Age of heifers at first service (months) 15.5 53
Age of heifers at first calving (months) 24.5 52
Percent of lactating cows open greater 19.6 48
than 150 days in milk
Average number of days open 106.8 46
Voluntary waiting period (days) 63 51
Average number of days to first service 70.7 51
Average calving interval (months) 13.1 50
Average length of dry period (days) 54.3 58
Percent of farms using Al to breed cows 78 58
Percent of farms using Alto breed 64 57
heifers
Average percent of cows being bred AI 87 45
of farms using AI
Average percent of heifers being bred 89 37
A1 of farms using AI
Average cost per straw of semen used 17.30 43
on cows ($)
Average cost per straw of semen used 18.83 35
on heifers ($)

 

I9

 

A series of in-depth questions regarding AI usage were asked in the survey,
including whether AI is used on cows and/or heifers, and the average cost of semen used.
Overall, 78% and 64% of farms surveyed indicated that Al was used to breed cows and
heifers, respectively, for at least some services. Caraviello et al. (2006), in their survey of
dairy farms, also sought to determine the extent of AI use and found that 58 of the 103
herds surveyed, or 56%, used solely AI. Zwald (2003), in comparing 14,500 herds, found
that approximately half of the herds used a bull for at least some services.

In order to assess the reasons for not exclusively using Al to service cows and
heifers, respondents were asked to select reasons for not using only AI. For cows, the
most frequently cited reason was a lack of labor for heat detection and to perform A1,
with 35% of total responses (34 total responses for cows). Additionally, semen costs
were cited in 24% of responses, other reasons not listed were cited in 20% of responses,
using a clean-up bull was cited in 12% of responses, and lacking handling facilities was
cited in 9% of responses as to why AI alone was not used to breed cows. Similar to
cows, the most prominent reason reported for not using solely Al on heifers was a lack of
labor for heat detection and to perform A1, with 29% of total responses (34 total
responses for heifers). Heifer responses differed from cows in that 26% of responses
cited a lack of handling facilities, 21% cited other reasons not listed, 12% cited semen
costs, and 12% cited the use of a clean-up bull as the reasons that solely Al was not used
to breed heifers.

Several respondents for both cows and heifers indicated that there were reasons
for not using solely AI beyond those listed in the survey. Some of those other reasons

given in survey responses were increased convenience with natural service, poor HDR,

20

seasonal calving schedules that required a tight breeding window which is better
accomplished through natural service, poor CR with AI, that natural service allows longer
seasons for pasture use, and that natural service yielded better results during summer heat
stress.

Accurate and efficient heat detection is crucial in the management of an AI-based
reproductive program and is necessary for managing a profitable dairy farm (N ebel and
Jobst, 1998). Nebel and Jobst (1998) found that accurately and efficiently performing
heat detection was the major limiting factor for reproductive performance on many dairy
farms. Survey respondents were asked to select from a multiple choice list all heat
detection methods used in 2005 in cows or heifers. Heat detection methods provided in
the survey included visual heat detection without aides, passive mount detectors, and
electronic heat detection aides. Respondents were also invited to add any additional
methods that they employed that were not listed as options in the survey. Passive mount
detectors listed for selection in the survey included Kamar Heatrnount Detectors
(Steamboat Springs, CO), chin ball markers, tail chalking or crayon, and a section for
other passive mount detectors. Electronic aides listed in the survey included HeatWatch
Estrus Detection System (CowChips, LLC, Denver, CO), pedometers, AfiAct System and
associated herd management software (SAE Afikim, Kibbutz Afikim, Israel), or other
electronic aided heat detection methods. Multiple responses were allowed for both cows
and heifers, and the total number of responses was 82 for cows and 53 for heifers. A

summary of the heat detection methods used by respondents is presented in Table 2d.

21

Table 2 (1. Summary of survey results to questions regarding heat detection methods
used for cows and heifers (number of responses = 82 and 52 for cows and heifers,
respectively)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Type of heat detection method Cows I Heifers
Percent of
responses
Visual Visual heat detection without aides 62 74
heat
detection
Passive Karnar HeatMount Detectors 5 2
mount Chin ball markers . O 0
detectors Tall chalking, crayon, or pamt 20 11
Other passive mount detectors 2 6
. HeatWatch 1 2
Electronic
aided heat Pedometers . 0 0
detection AfiAct and assocrated herd management program 1 0
Other electronic aided heat detectionjrogram 0 0
Other Other method not categorized above 9 5

 

Visual heat detection without the use of aids was the most prominent heat
detection method used for both cows and heifers. Also for both cows and heifers, tail
chalking, crayons, or paint was the second most common method of heat detection
employed. Caraviello et a1. (2006) also found that tail chalk was the most common heat
detection aid out of their list of tail chalk, pedometers, pressure patches, and other, with
tail chalk receiving 60 of the 80 responses regarding heat detection aides used.

If visual heat detection was being used in either cows or heifers, respondents were
asked to provide additional information on the times per day animals were observed, for
how long animals were observed each time, and who was responsible for heat detection.
Of those farms reporting the use of visual heat detection in cows, on average 78% of the
cows on those operations were bred solely by visual heat detection. Of those farms

reporting the use of visual heat detection in heifers, on average 90% of the heifers on

22

those operations were bred solely by visual heat detection. The average overall HDR
reported for cows was 52%.

On average, cows and heifers were observed for estrus 3 and 2.2 times per day,
respectively. These observation frequencies for heat detection were similar to those
found by Caraviello et al. (2006) where cows were reportedly checked for estrus 2.8
times per day on weekdays and 2.5 times per day on weekends. In addition, Stevenson
(2003) indicated that in a survey of top dairy herds as measured by yearly rolling herd
averages, cows were observed for estrus 3.1 times per day, on average, and that this was
likely responsible for successful AI breeding. The most commonly reported times for
heat detection were moving cows, pre-milking and post-milking, and while feeding. Of
the 31 farms reporting heat detection times the average times spent observing cows and
heifers were 43 and 19.5 minutes per observation, respectively. Compared to previous
survey results by Caraviello et al. (2006) which indicated cows were observed for 27
minutes on weekdays and 25 minutes on weekends per observation, cows were reportedly
observed for longer and heifers were observed for a shorter time period.

Questions regarding who was responsible for heat detection did not provide
multiple choices for respondents, but allowed respondents to indicate people as they
wished. Job levels were then grouped and frequencies were calculated. Of the 42 total
responses, the person most commonly responsible for heat detection was the owner with
55% of responses, followed by a shared responsibility by all employees which received
26% of responses, herdsman with 17% of responses, and milkers with 2% of responses.

The person responsible for heat detection will likely affect the true cost of a heat

23

detection program as the owner or herdsman will likely have a higher labor cost than
other farm employees.

Separate responses regarding synchronization programs were invited for cows and
heifers to allow for different management programs. In total, 56 and 45 responses were
received for whether any synchronization program was used for cows and heifers in
2005, respectively. Synchronization programs were used proportionately more in cows
than heifers, with 45% of responses indicating the use of some synchronization program
in cows versus only 27% in heifers. In comparison to previous survey analysis by
Caraviello et al. (2006) which found that the majority of herds in their survey used
hormonal synchronization or TAI programs, a significantly smaller proportion of survey
respondents used synchronization programs. Possible explanations for the smaller
proportion of herds having used synchronization programs are the differences in average
herd size between the two surveys or even the differing sampling methods. Given that
prior studies have surveyed farms associated with a given organization, the type of
organization that a survey is associated with may affect the willingness of survey
participants to use certain technologies.

Respondents who indicated no synchronization programs were used were asked to
select reasons why they did not use a synchronization program from a multiple choice
list. Possible reasons provided for cows or heifers included the expense of
synchronization programs, manager or breeder preference to breed cows off of visual
heat detection, inadequate facilities to restrain cows for injections, lack of management

time to manage a synchronization program, not being convinced of the benefits of

24

synchronization, poor previous CR to TAI, and other. A summary of the responses to

why synchronization programs were not used is provided in Table 2e.

Table 2 e. Summary of survey results regarding reasons why synchronization
programs were not used for cows or heifers (number of responses = 25 and 43 for
cows and heifers, respectively)

 

 

 

 

 

 

 

 

 

 

Reason Cows I Heifers
Percent of resmnses
Synchronization protocols too expensive 16 12
Prefer to breed to visually detected estrus 28 24
Inadequate handling facilities 4 21
Lack of management time 8 9
Not convinced of benefits of synchronization 16 12
Poor previous conception rate to timed Al on 8 3
farm
Other 20 21

 

 

 

 

In order to determine which synchronization programs were used, a list of
common programs was provided and respondents were asked to select any programs they
used on cows or heifers or to provide information on any other protocols they have used,
and explain what proportion of the herd each program was used on. Separate answers
were encouraged for heifers and cows to allow for differing management programs.
More responses for programs used were received than the number of farms reporting
having used a synchronization program as several farms used multiple synchronization
programs. Given the small sample of farms which used synchronization programs for
heifers in 2005, the proportion of the herd having used each program was not calculated
due to insufficient number of farms using each program to make the proportion of the
herd valuable. Table 2f provides a summary of the responses for cows and heifers related

to synchronization programs.

25

Table 2 f. Summary of survey responses regarding synchronization programs used
(number of responses = 53 and 15 for cows and heifers, respectively)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Question Cows Heifers Percent of cows
on which the
program was used
Percent of responses
Ovsynch 38 33 42
Presynch 13 7 72
Cosynch 6 0 57
Heatsynch 2 0 10
Controlled internal drug-releasing 19 20 7
intravaginal insert (CIDR) with
PGFZa
Targeted breeding protocol 6 20 63
Single PGFZa injection (with AI 13 20
upon detected estrus)
Single PGFZa injection with timed 2 0
AI
Other 2 0 14

 

Farms were also asked to provide cost information for any treatments used in
cows or heifers. Costs per dose were collected and ranged significantly across farms in
the survey. For example, PGF2a costs per dose ranged from $1.25 to $6.00 per dose.
Given the large variation in costs per dose reported, the costs of hormones available to a
specific farm may indeed be different than those available to another, and such
differences may alter decisions made regarding reproductive management programs.
Overall, the average costs per dose reported for 2005 for all respondents were $3.59 per
dose for GnRH, $2.52 per dose for PGF2a, and $9.22 per CIDR. When farms were sorted
into groups of 100 cows or less, 101 to 200 cows, and more than 200 cows, the costs per
injection varied significantly across farm sizes. Costs for GnRH per dose were $4.49,
$3.13, and $2.70 per dose for those size categories described above. PGFZa costs by

those same size categories were $3.10, $2.13, and $2.10 per dose.

26

In order to more completely assess a farm’s reproductive management program
and decisions, respondents were also asked the amount of time needed per cow to give a
single injection and the person responsible for giving synchronization program related
injections. Time required for injections and the facilities used for injections varied
considerably across farms, ranging from 17 seconds where shots are given to cows
already in the milking parlor to 10 minutes where heifers must be caught in a freestall and
put into a headlocks one at a time. Given the wide variation in facilities used and the
time it takes to give an injection, ideal reproductive management programs will be
different for farms with varying circumstances. Across the wide variety of facilities used
for cows and heifers on the 26 farms responding to this question the average time taken to
give an injection was 2.1 minutes. Twenty-seven responses were received regarding the
person responsible for injections related to synchronization programs and were
categorized. The person on the dairy responsible for synchronization-related injections
the majority of the time was the owner with 59% of responses. Following the owner, in
order of frequency were the herdsman or herd manager, milker labor, AI technicians and
farm family members. The person responsible for giving injections affects the cost of a
synchronization program due to differences in costs of labor across farm workers.

Respondents were asked to comment on their most recent major reproductive
management change and report in what year it took place. Several farms indicated the
initiation of a synchronization program or moving to AI from natural service as their
latest reproductive management change. In light of recent developments in reproductive
technologies for on—farm use, such as ultrasound for pregnancy detection and sex-sorted

semen allowing altered sex ratios, perhaps the most surprising changes reported were the

27

several farms that reported a departure from the use of reproductive technologies.
Examples included giving up a synchronization program in favor of unaided visual heat
detection or moving from an AI based breeding system in favor of natural breeding. AI
has been said to have been the most readily accepted technology on the dairy farm, with
the exception of the milking machine (Stevenson, 2003). Even so several reasons were
given for using bulls over using AI, ranging from costs associated with Al to the belief
that the bull was better at heat detection and yields a higher CR. Other changes reported
included increased time devoted to visual heat detection rather than relying upon heat
detection aides, switching AI technicians, or ceasing use of hormones for
synchronization.

With several farms reporting their most recent change being reverting to natural
service it should be highlighted that there are several concerns associated with keeping
dairy bulls on the farm. Costs associated with the bull can range from daily maintenance
costs to costs for diseases spread by bulls throughout the herd. In addition, the bull-to-
female ratio on the particular farm and the number of pregnancies produced by the bull
each year will affect the cost per pregnancy associated with using natural service (Fricke
and Niles, 2003). Increased HDR and CR are common reasons for maintaining a bull
within the dairy herd, but without testing, the fertility of the on—farm bull is unknown. In
addition, Schutz (Purdue University, September 2007, personal communication, e-mail)
highlights that while using a bull for natural service may seem like a good way to
increase CR, in reality additional factors may be introduced that are difficult to monitor,
such as infertility, reduced breeding success due to sperm abnormalities, or decreased

libido (particularly during hot weather).

28

F ricke and Niles (2003) warn that in many cases, costs of maintaining bulls can
approach or exceed AI related costs without consideration of the long-term genetic
advantages of using AI. Overton (2005) states that the additional lost opportunity cost of
the room which could be used to house more lactating cows if the bulls were removed
should also be recognized. Perhaps the largest factor to consider when assessing the
decision to use bulls for natural service is the safety of those employees, family members,
or others who may visit the farm. Schutz (Purdue University, September 2007, personal
communication, e-mail) pointed out the often overlooked danger for farm employees
unfamiliar with working with dairy bulls, particularly on expanding dairy farms where
bulls may be combined with employees with limited experience in working with cattle.

Of those 26 farms detailing their most recent change in response to the survey,
77% of farms latest reproductive management change took place between 2000 and 2005.
With such a high percentage of farms making recent reproductive management changes,
increased volumes of research revolving around reproductive management programs are
warranted by industry action.

Johnson (2005) applied the principle of diminishing marginal returns to dairy
farm inputs to highlight that given decreasing marginal benefits of a costly input, the best
economic returns are nearly always found at some level of that input which is lower than
the level that gives the highest output. Further, Johnson (2005) explains that the higher
the cost of the input relative to the value of the output, the less of that input you will
invest in order to be profitable. These economic principles, when applied to dairy farm
inputs which produce decreasing marginal effects, imply that farmers must select levels

of inputs which maximize their profitability. Further, the levels of inputs chosen to

29

maximize profitability will almost certainly not be those levels which maximize
production. For example, seeking an 80% HDR through visual heat detection may be
attainable, but the time and money invested in doing so may be difficult to justify
economically.

Prior reproductive performance was important in assessing which reproductive
management program was the optimal program for a given farm. Farms which have
already experienced success in their reproductive management programs will have less
economic incentive to pursue reproductive management changes due to decreased
marginal returns from increased reproductive performance. Overall, farms which have
experienced success with their current programs stand to gain less from reproductive
improvements than farms which have faced reproductive performance challenges.

Other farm management factors, perhaps not as easily quantified as labor or
treatment costs, exist when decisions regarding reproductive management are made. For
example, farm managers may prefer to work in a particular environment, or to work with
only family labor versus expanding the dairy and managing several employees. Human
resource management challenges were identified by Caraviello et al. (2006) in their
survey of dairy herd managers, where managers identified finding good employees as the
greatest labor challenge, followed by training and supervising employees. Managers of
smaller dairies often supply much of the day-to-day labor and focus on herd or crop
management activities. Increased dairy farm size leads to focus being placed on new
areas, such as financial requirements and implications of expansion; hiring, training,
evaluating and retaining employees; sourcing adequate inputs (e.g., land, feed,

replacements); meeting environmental and zoning regulations; and managing public

30

relations (Hadley et al., 2002). Managers who enjoy day-to-day hands-on management
of the dairy herd or cropping operation may seek ways other than expansion to increase
profitability. These management preferences and farm management goals must be taken
into account when determining farm-specific optimal reproductive management
programs.

Use of synchronization programs aimed at decreasing the time, and therefore
expense, devoted to heat detection aim to either shorten the time during which cows or
heifers must be observed, or eliminate the need for heat detection completely with TAI.
Timed ovulation allows for breeding cows by appointment, thereby eliminating the need
for heat detection, and has similar pregnancy outcomes to those obtained through AI
performed after heat detection (Pursely et al., 1997). The benefits of adopting a program
designed to decrease heat detection duties will be dependent on the HDR and efficiency
achieved on the dairy previously. Given the farm-specific aspects of reproductive
management decisions the cost effectiveness of using a systematic breeding program
must be assessed for each herd to decide if such a management program is the optimal
choice (Nebel and Jobst, 1998).

Focusing on reproductive programs using AI, program costs for visual heat
detection and synchronization protocols were calculated and sensitivity analyses
performed. Using the base scenario described in Materials and Methods with a group
size of 100 cows, costs for visual heat detection without aides and synchronization via
Ovsynch were focused upon. Costs of AI programs are presented in Figure 1. Program
costs for four reproductive management programs are displayed together to allow

comparison across increasing labor costs. As labor costs increased, holding all other

31

parameters constant, different reproductive management programs became the optimal
program to achieve the 90% cumulative probability of pregnancy. The differences in the
slopes of the lines for the HDR achieved through visual heat detection versus the line for
Ovsynch indicate the differing relative labor intensities of the programs. Visual heat
detection requires more labor hours per cow and was more sensitive to increased labor
costs than the administration of a synchronization program.

The effect of labor efficiency in heat detection is highlighted in Figure 2a by
comparing the costs associated with achieving a 65% HDR with 2.15 labor hours per day
invested versus the 65% HDR with 2.6 labor hours per day invested. Farms with higher
levels of labor efficiency in heat detection have lower per cow costs than those with less
efficient heat detection. The more efficiently the farm was able to detect heats visually,
the higher the cost of labor before the Ovsynch program becomes the least cost program.
Simply stated, the better the current performance in visual heat detection, the higher the
labor cost must be before synchronization with Ovsynch becomes a lower cost program
than visual heat detection. This supports the contention that current farm-level
reproductive performance is important in assessing the lowest cost program for a

particular farm.

32

Figure 2 a. Artificial insemination program costs: sensitivity to heat detection rates
and heat detection efficiency

 

Artificial Insemination Program Costs: Sensitivity to Heat Detection Rates
and Heat Detection Efficiency

 

 

 

 

 

 

 

 

 

3

O

U

i-

2. + HDR 52% with

g 2.15 labor hours per day

U + HDR 65% with

g 2.15 labor hours per day

30 + HDR 65% with

5 2.6 labor hours per day
+ Ovsynch with 2.1

Minutes per Injection

 

 

Labor Cost per Hour

 

 

 

 

Given the base scenario depicted above, with a HDR of 52% achieved through
2.15 labor hours per day of heat detection being compared to the Ovsynch program, the
two programs have the same cost at a labor cost of $5.39 per hour. Therefore, at any
labor cost greater than $5.39 per hour, the Ovsynch program is less costly than visual
heat detection. If a 65% HDR requires 2.6 labor hours per day of heat detection to
achieve, the labor cost below which visual heat detection was less costly is $7.74 per
hour. With greater efficiency in detecting heats, where a 65% HDR was achieved
through 2.15 hours of labor per day, the labor cost below which visual heat detection was
less costly than Ovsynch is $9.55 per hour.

Another key parameter in determining and comparing program costs is the time
required to give an injection in the synchronization protocol. Program costs were
calculated for the Ovsynch program with 2.1, 4.2, and 6.3 minutes per shot. Program
costs, compared with a visual heat detection program in which a HDR of 65% was

achieved with 2.15 labor hours per day are presented in Figure 2b. Figure 2b clearly

33

displays that the visual heat detection program was the least costly at relatively low labor
cost, and the cost of labor at which the synchronization program becomes the lower cost
program depends on the time required per injection. Using a visual heat detection
program where a 65% HDR was achieved through 2.15 labor hours per day, the Ovsynch
protocol becomes the lower cost program at a labor cost of $9.55, $10.73, and $12.26 for
2.1, 4.2, and 6.3 minutes required per injection, respectively. The longer it took for a
single injection to be administered, the more sensitive the program was to the cost of
labor, and the higher the labor cost must be before the synchronization program is the

lower cost program in comparison to visual heat detection.

Figure 2 b. Artificial insemination program costs: sensitivity of Ovsynch
synchronization program to time per injection

 

 

 

 

 

 

 

 

 

 

 

 

Artificial Insemination Program Costs: Senstivity of Ovsynch
Synchronization Program to Time per Injection
g 3200-00 “ n/‘ + HDR 65% with
U /n/ 2.15 labor hours
:3. $180.00 r -—— a? */./8‘~"————* per da
D- / y .
:3 $160.00 4 —— a + Ovsynch wrth 2.1
5 _‘ .. Minutes per
5 $140.00 ‘ ‘-“""' "' F! ‘_ _ "' _ ’ Injection
is.” $120.00 *l/e g _ + Ovsynch With 4.2
2 Minutes per
“ $100.00 . I I r I . . I . . . . . . Injection
+ Ovsynch with 6.3
409° $3.09“ $9“ 09“ g“ 4,9“ 69° “99° Minutes per
‘9 ‘5' ‘5' ‘3 ‘5 ‘9 Injection
Labor Cost per Hour

 

2.4 Conclusion

 

 

This analysis built upon prior reproductive management studies and dairy industry

surveys to inform the economic analyses of various reproductive management programs.

34

Costs associated with reproductive technologies, and with a given level of reproductive
performance, vary considerably across farms, illustrating the need for farm-specific costs
in the analysis regarding which reproductive management program is best for a given
farm operation.

Reproductive program costs were found to be highly sensitive to on-farm labor
costs. As labor costs increased, holding all other parameters constant, differing
reproductive management programs became the lowest cost program of achieving the
performance parameter of a 90% cumulative probability of pregnancy. In addition,
different types of reproductive management programs had different relative sensitivities
to costs, such as on-farm labor costs. For example, visual heat detection requires more
labor hours per cow and therefore is more sensitive to increasing labor costs than a
synchronization program. Current farm-level reproductive performance was also found
to be important in assessing the lowest cost program. Farms which have obtained high
levels of visual heat detection efficiency, for example, have less incentive to adopt a
synchronization program than those farms with less efficient visual heat detection.

Overall, reproductive management programs differ across farms due to varying
on-farm costs, facilities, farm goals and values, and management styles. These factors, in
addition to labor availability, labor costs, ease of cattle handling, and previous levels of
reproductive performance helped to determine why different farms select different
reproductive management techniques and technologies when seeking to maximize their

profitability through reproductive performance.

35

CHAPTER 3: DECISION SUPPORT FOR REPRODUCTIVE MANAGEMENT
TECHNOLOGY AND PROGRAM ADOPTION ON U.S. COMMERICAL DAIRY
FARMS

3.1 Introduction

Economic analysis of reproductive management programs and technologies is
motivated by the fact that reproductive performance and efficiency of dairy cattle affects
dairy herd profitability through milk yields and revenues, numbers of days open, and
potentially expensive culling repercussions of poor reproductive performance. In short,
reproductive performance and efficiency are critical drivers in determining the
profitability of the dairy herd.

Efficient and accurate heat detection is crucial to achieve optimal management of
individual cows and the major limiting factor to reproductive performance is estrus
detection (Nebel and Jobst, 1998). In addition to challenges with detecting cattle in
estrus, first artificial insemination (AI) service conception rates (CR) for cows have
fallen substantially from 60% to 40% since AI has been practiced in the Unites States
(N ebel, 2002). Movement towards increased herd sizes and higher producing cows has
been associated with decreases in reproductive efficiency (Lucy, 2001). Combining the
challenges of identifying cattle in heat, decreasing CR, and the need for management to
adjust techniques and methods as herd sizes increase has led to industry-wide challenges
with achieving optimal reproductive performance. The decline in reproductive
performance is likely partially explained by physiological adaptations to high milk
production (Lucy, 2001), although other factors are likely responsible as well. In
response to challenges associated with achieving reproductive efficiency, several

reproductive management programs have been developed to synchronize estrus or

36

ovulation, thereby greatly reducing or entirely eliminating the need for visual estrus
detection on dairy farms. The costs associated with visual heat detection are heavily
dependent on the efficiency with which on-farm labor was able to detect heats, although
labor costs and other farm factors will influence the program costs associated with visual
heat detection as well.

Dairy farmers can benefit from decision-support which aids them in selecting the
optimal available reproductive management program or technology for their operation.
“Rules of thumb” can aid producers in deciding which programs may potentially work on
their operation, although on-farm support can aid farmers in selecting the optimal
program for their situation specifically. Producers must remember that program or
technology adoption decisions must be made subject to the available technology set, skill
sets of managers and other farm labor, previous levels of reproductive performance, and
farm goals and values. A decision support-tool was developed for on-farm use to aid
dairy farmers in decision making surrounding economically optimal choices for
reproductive management technologies and programs. Dairy farmers, veterinarians,
extension agents, and consultants in the dairy industry could all benefit from a user-
friendly decision support tool capable of providing economic analyses of reproductive
management programs and technologies.

The decision-tool developed in this analysis is based upon a model which was
developed to calculate the expected value associated with a given reproductive program,
subject to user-inputted farm-specific values. Program values are calculated by
determining the expected revenues and expected costs. Program costs are calculated

subject to user-defined cutoff points after which the cow is no longer bred, but is assessed

37

at the retention pay-off (RPO) of an open cow at that point in time. Technologies and
programs which can be analyzed using the on-farm decision tool include synchronization
programs (e.g., Ovsynch, Cosynch, Targeted Breeding ProtocolTM (Pharmacia-Upjohn,
Kalamazoo, MI), Presynch), CIDRS, passive visual heat detection aides, and electronic
heat detection aides. Common passive estrus detection aides include Kamar Heatrnount
Detectors (Steamboat Springs, CO), chin ball markers, and tail chalk or crayon.
Common electronic aides include HeatWatch Estrus Detection System (CowChips, LLC,
Denver, CO), pedometers, AfiAct System and associated herd management software
(SAE Afikim, Kibbutz Afikim, Israel). The extent to which heat detection aides improve
reproductive performance will depend on farm-specific characteristics and farm
management abilities and knowledge. The tool was designed to be flexible in allowing
user-specified programs to be entered as sequences of shots or combinations of heat
detection aids, thereby allowing producers to alter programs or to assess new programs as
they become available.

Several Spreadsheet-based models exist to aid producers in decision making
regarding reproductive programs. Groenendaal et al. (2004) constructed a spreadsheet
based model to aid in optimal breeding and replacement decisions on the dairy farm. The
results of Groenendaal et al. (2004) indicated that the costs per additional day open
ranged from $0 to $3.00 in which heifers were available as replacements (using the
typical input parameters for Pennsylvania); the results were dependent on many factors,
four of which were availability of replacement heifers, lactation number, milk production
levels (of individual cow and herdmates) and the calving interval. Meadows et al. (2004)

developed a Spreadsheet based model to allow comparisons of scenarios and to determine

38

the economic effect of varying reproductive performance in dairy herds. Meadows et al.
(2004) highlight that “. .. inefficient production becomes marginally more costly to
producers as performance declines and warrants increased attention.” In short, the better
the present reproductive performance of a herd, the less economic incentive there is to
improve that performance (Meadows et al., 2004).

Flexibility to incorporate varying farm systems, on-farm costs, and labor
scenarios is important in creating a tool that can be used for decision making across
various farm operations. A difficult balance must be struck in spreadsheet-based models
intended to be used on-farm as decision support between incorporating the complexity
needed to adequately model the problem while retaining simplicity to make the model
user-fiiendly and easily understood. Tools which allow reproductive program protocols
to be parameterized by the user, in terms of what injections are given within a program,
will be better suited to adapt to new programs as they are developed.

The objective of this paper is to describe a user-friendly spreadsheet tool capable
of farm-specific analysis of values of reproductive management programs which use Al.
The decision support tool described in this paper is capable of assessing any user-
specified synchronization program through entering program protocols and the farm-
specific costs for each portion of that protocol. Reproductive performance, or expected
reproductive performance, for a given farm operation under each program can be entered,
and sensitivity analysis to key parameters is easily completed. Further, the tool was
designed to allow flexibility by allowing the farm to specify CR, AI submission rate,
labor costs, and other reproductive parameters and associated costs. The goal of the tool

itself was to provide decision support to dairy farmers as they make reproductive

39

technology and program adoption decisions for various groups of cattle on their dairy
farm. Decisions made for individual groups are possible within the tool as costs per cow

are adjusted according to the number of cattle in the group.

3.2 Materials and methods
Tool description and development

The decision-tool was developed in Excel (Microsoft, Seattle, WA) to determine
farm-specific or individual group-specific costs and values associated with achieving
various levels of reproductive performance through the use of various reproductive
management programs. All reproductive management decisions made within the tool are
assumed to use AI as the sole breeding method.

The decision-tool allows user-specified inputs for key farm values and costs
associated with reproductive management programs and technologies. The inputs
allowed into the decision-tool are explained in the sections that follow. AI service cutoff
criteria are necessary in order to compare values across reproductive management
programs. Values of reproductive programs are assessed subject to two possible cutoff
criteria for breeding, namely a herd-manager specified number of AI before a cow will no
longer be bred or a herd-manager specified cumulative probability of pregnancy. In order
to calculate the cumulative probability of pregnancy, the pregnancy rate (PR) is
estimated by:

PR = CR * SUBRATE,
where

PR = pregnancy rate,

40

CR = conception rate, and

SUBRATE = AI submission rate.

Using the estimation above, the cumulative probability of pregnancy can be
calculated after each AI. Clearly the PR is dependent on the CR and AI submission rate,
and therefore changes in the CR or Al submission rate will affect how many services are
necessary to achieve a specified cumulative probability of pregnancy. In order to
compare values across synchronization programs the values must be assessed subject to
achieving a specified level of reproductive success and discounted to a specific point in
time. If the decision is being assessed by using a cutoff number of AI after which the
cow will no longer be bred, then the CR and AI submission rate will affect the remaining
probability that the cow is not pregnant (or remains open) after that predetermined cut-off
number of services. For instance, scenarios can be assessed in which cows are bred for a
predetermined number of AI. Costs can be extremely sensitive to the cutoff criteria
employed, and such sensitivities are assessed.

Using a scenario in which the CR was 35% and the AI submission rate was 50%,
the cumulative pregnancy rate is calculated in Table 3a for illustration. For comparison,
the cumulative pregnancy rate for a synchronization program scenario in which the AI
submission rate is 100% and the CR is 35% is shown in Table 3b. The breeding in which
90% cumulative probability of pregnancy is achieved is highlighted in both Tables 3a and

3b.

41

Table 3 a. Calculating cumulative probability of pregnancy (AI submission rate =
50%, CR= 38%)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Breeding Probability of Remaining Cumulative
pregnancy probability of probability of

being open pregnancy
1 0.18 0.83 0.18
2 0.14 0.68 0.32
3 0.12 0.56 0.44
4 0.10 0.46 0.54
5 0.08 0.38 0.62
6 0.07 0.32 0.68
7 0.06 0.26 0.74
8 0.05 0.21 0.79
9 0.04 0.18 0.82
10 0.03 0.15 0.85
11 0.03 0.12 0.88
12 0.02 0.10 0.90

 

 

 

 

 

Table 3 b. Calculating cumulative probability of pregnancy (AI submission rate =
100%, CR= 35%)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Breeding Probability of Remaining Cumulative
pregnancy probability of probability of

being open pregnancy
1 0.35 0.65 0.35
2 0.23 0.42 0.58
3 0.15 0.27 0.73
4 0.10 0.18 0.82
5 0.06 0.12 . 0.88
6 0.04 0.08 0.92
7 0.03 0.05 0.95
8 0.02 0.03 0.97
9 0.01 0.02 0.98
10 0.01 0.01 0.99
11 0.00 0.01 0.99
12 0.00 0.01 0.99

 

 

 

 

 

In order to allow the most accurate decisions possible for individual cattle
groups, herd managers are able to enter values for cow or heifer groups, thereby creating

specific analysis for cattle groups on the farm, rather than imposing the same values (or

42

program) on the entire herd. For example, given that group-specific input values were
available, the cost calculations are sensitive to the number of cows on a given program,
and the tool is able to determine the economically optimal prOgram for given groups of
cows while taking into account any economies or diseconomies of scale associated with
the number of cows managed under a specific reproductive management program. The
group size is most important when assessing programs using visual heat detection
because the hours of labor spent detecting heats for a group is spread over the number of
cattle which are watched at a single time. For example, if an hour a day is spent detecting
heats on a group of 50 cows, the per-cow costs associated with labor for heat detection
would be higher than if that same time was Spent watching a 100 cow group.

Time required to administer injections is an important component of cost and is a
function of facilities employed and skill level. Similarly, visual heat detection program
costs vary depending on hourly labor costs and efficiency with which heats are detected.
Reproductive program costs were calculated on a per cow basis to facilitate comparison
across programs and herd sizes. Heat detection program costs were adjusted to obtain a
per cow basis by dividing the cost of heat detection for a group of cows over the munber
of cows in the group. In calculating the program costs for both synchronization and
visual heat detection programs, the cost of labor to perform the program was assessed as
wage paid if the task was performed by paid labor or as opportunity cost of the labor if
unpaid labor was used. Opportunity cost is cost of having that labor participate in the
reproductive management tasks rather than in the next best alternative activity. Due to
the fact that many herds, especially smaller herds, utilize unpaid family or managerial

labor for their reproduction program, in order to assess the true costs associated with

43

various reproductive management programs a charge for the opportunity cost of unpaid
labor must be included.

While heat detection rate is the percentage of cows correctly detected in estrus in
a 21 days period, AI submission rate is the percent detected in heat and bred in that same
period. The cumulative probability of pregnancy was defined as the sum of PR from the
first AI through the most recent AI. In calculating number of services necessary to reach
a cumulative probability of pregnancy, the CR was assumed to decrease as Al number
increased. The second and third or later AI were assumed to be 90.7% and 81.4% of the
first AI CR, based on Cassell et al. (2001). The AI submission rate was held constant for
each program.

Once the predetermined cutoff was achieved, the cow’s value was included by
incorporating her retention pay-off (RPO). The RPO is the difference in total net returns
from keeping the cow in the herd versus culling and replacing her immediately. The
RPO is defined as the total additional profit expected from attempting to keep the cow
until her optimal age compared to immediately removing her from the herd and replacing
her (taking into account changes in involuntary culling); the higher the RPO, the more
valuable the animal and, thus, the larger the loss if the cow is culled at that time
(Groenendaal et al., 2004). The RPO values used were obtained using DairyVIP©
version 1.1 (De Vries, 2006). Input values used to obtain the RPO for pregnant and open
cows were from De Vries (2006) unless otherwise specified.

For each period, the RPO of a cow which conceived in that period was multiplied
by the CR for that period and discounted. The probability that the cow remained open

afier the last breeding period was multiplied by the RPO of an open cow in the period in

44

which the cutoff criteria was reached. The cost per breeding for calculating the RPO was
set at zero because breeding costs were accounted for in calculating the net present value
(N PV) of the breeding programs. The RPO values included in the model were greater
than zero and values that were negative were evaluated at zero. Feed and yardage of
$2.00 per day was charged in the calculating of the breeding program cost for each period
that the cow remained open after her first AI. An annual discount rate of 9% was used
for all calculations, as an entry for the RPO calculation as well as for calculating the NPV
of the breeding programs (Wolf et al. 2002). Cows were assumed to be bred beginning at
their third month in lactation and were bred once during every eligible period until the
cutoff criteria was reached, which in the example cases used here was a cut-off criteria of
6 AI. Eligible periods for breeding differed based upon the program in which cows were
being bred. Cows bred with visual heat detection were assumed to have breeding periods
of 21 days. Cows on a synchronization program were assumed to begin
resynchronization upon a nonpregnant diagnosis at either 26 or 33 days after TAI.

Heat detection program costs, assessed on a per period per cow basis were

calculated as follows:

PROGHD = [(((TIME * OBS) * PER) * LABORHD)/COWS] + AID + (Cl * SUBRATE),
Where,

PROGHD = Heat detection program costs per cow per period,

TIME = Heat detection minutes per observation for a group of cows,
OBS = Number of times the group is observed per day,

PER = Number of days in a single breeding period,

45

LABORHD = Cost of labor (in dollars per minute) to perform heat detection,

COWS = Number of cows in the group,
AID = Per period cost per cow of heat detection aid utilized,
CI = Cost of an artificial insemination, and

SUBRATE = AI submission rate.

Synchronization program costs, assessed on a per period per cow basis were

calculated as follows:

PROGSynch = (PGnRH * XGnRH)+(PPGF2a * XPGF2a) ‘l' (MIN * INJ * Labor Inj) + (C1

* SUBRATE),

where,

PROGsynCh = Synchronization program costs per cow per breeding period,
PGnRH = Cost of GnRH per injection,

XGnRH = Number of GnRH injections administered,

PPGFZQ = Cost of PGFZa per injection,

ngp2a = Number of PGFZa injections administered,

MIN = Number of minutes to give a single injection,

INJ = Total number of injections in the series,
LABORInj = Cost of labor (in dollars per minute) to give injections,

CI = Cost of an artificial insemination, and

SUBRATE = AI submission rate.

46

The formula used to calculate the expected NPV of each program, with NCFT for

the last time period including the RPO values:

NCFt = (CRt * RPOPREGt - (1 - PRcum)* Feed — PROGj) for t = 0,. . .,T-1, and
NCFT = (CRT * RPOPREGT - (1 - PRcum) * Feed — PROGj + (1 - PRcum) *
RPOOPENT) for t=T, where,

NCFt = Net cash flow in period t,

PROG = Program costs for either synchronization or heat detection program per

cow per period,

CRt = Current AI CR,

RPOPREGt = Retention payoff in time t for a pregnant cow,

PRQum = Cumulative probability of pregnancy, defined as sum of PR from 0
through the current AI, in time t, (e.g., PRcum, 3 = CR0 + (1 - CR0) * CR1 + (1 - CR0) *

(1 -CRI) * CR2 ).

Feed = Per period feed and yardage cost of nonpregnant cow = ($2.00 per day *
days per period),

PROGj = Cost of program j, where j is the reproductive program used on the cow,
RPOOPENT = Retention payoff in time T for an open cow, and

RPOPREGT = Retention payoff in time T for a pregnant cow.

47

The NPV of the strategy was the sum from time period 0 through T, in which the
final time period, T, was determined by a predetermined cutoff criteria, such as the

maximum number of AI:

 

T 1
NPV = Z * NCF .
t= 0j(1+ r)t tj
where r is the discount rate and NCF, is the net cash flow in period t.

Sensitivity analyses were performed for differing costs of labor, minutes required
per injection, and labor efficiency of heat detection, holding all other variables constant.
Further, sensitivity analysis was conducted for cows of different lactations. Heat
detection labor efficiency was altered by holding the AI submission rate constant while
changing the labor hours required to achieve that performance. Efficiency in giving
injections was altered by adjusting labor minutes required per single injection.

The synchronization protocol chosen for use in the example was Ovsynch.
Ovsynch consists of (1) an injection of GnRH, (2) an injection of PGFZa 7 days later, (3)
a second GnRH injection 48 hours later, and (4) timed breeding 24 hours after the second
GnRH injection (Pursley et al., 1995). In order to resynchronize cows not conceiving to
the first AI, all cows were assumed to receive a GnRH injection at a nonpregnant
diagnosis (assumed to be 33 days post Ovsynch TAI), a PGFZa injection 40 days after the
initial Ovsynch TAI, and a second GnRH injection and TAI approximately 42 days after
the initial Ovsynch TAI. For comparison, resynchronization beginning 26 days after
Ovsynch TAI was assessed. Fricke et al. (2003) compared these resynchronization

programs to other programs with altered timing to beginning of resynchronization. The

48

assumptions described here were used to create baseline farm scenarios. In this analysis

scenarios were assessed in which cows were bred for 6 AI.

Input parameters for tool

General farm information that is required for input into the tool include the
number of cows in the group, the labor costs for visual heat detection, the labor costs for
administering shots, the feed and yardage costs to maintain an open breeding-age heifer,
and the annual discount rate used to compute the expected net present value of breeding
strategies. The farm-specific costs and revenues assessed in this analysis can be
expanded to include altering the user-specified inputs in DairyVIP© version 1.1 (De
Vries, 2006) where values such as average heifer calf values, replacement heifer costs,
and milk prices can be changed. AI costs which are not program specific can be entered
into the tool directly, by specifying breeding costs of zero dollars in the RPO calculation
and deducting AI breeding costs directly in the Excel spreadsheets constructed. Costs
associated with AI which are not program specific would include the semen costs per
straw and the costs associated with the insemination itself. If AI is performed by an
outside inseminator, a per-AI fee is entered. If AI is performed by farm labor the labor
costs per hour and the minutes of labor required per AI are entered. The general formula

used to calculate the cost of an AI is as follows:

MinsB d
Cost per AI = Straw + Fee + ___Lee_ * Labor ,
6O Breed

where
Straw = Cost of semen per straw,

Fee = A1 fee for outside inseminator for single breeding,

49

MinSBreed = Minutes of on-farm labor required for a single breeding, and

LaborBreed = Cost per hour of labor that performs Al on the farm.

Different costs for labor associated with insemination (if done by farm labor),
heat detection, and administering shots are able to be entered. The difference in labor
costs for each of these tasks allows farms to accurately account for different skill levels,
and therefore pay levels, of labor associated with each of these tasks. For example,
feeders or milkers on the farm may be responsible for heat detection duties, while
administering shots for a synchronization program is done by the herdsman or manager.
Additionally, heat detection performed by the manager of the herd will be more costly
than if other employees completed the heat detection. In short, given the sensitivity of
program costs to labor costs, the person on the farm who is performing a given task may
affect which program is economically optimal for a given farm operation.

To evaluate synchronization programs, users must specify their expected CR and
AI submission rate. Note that if ovulation synchronization is used, the AI submission
rate will likely be approaching 100%, because nearly all cows which are on the program
will be submitted for AI, barring any unusual circumstances. A key value that must be
entered when considering synchronization programs is the number of minutes it takes to
administer a shot to a cow in the group to be evaluated. Several factors affect the minutes
required per shot, and the total time is likely a function of the ease of handling cattle
within the facilities. Again, individual group evaluations are helpful in determining
optimal programs for a given group of cattle on the farm when facilities, or ease of
handling, in one group is different from another group. For example, when selecting a

program for heifers which are housed away from the home farm on a bedded pack

50'

without adequate headlocks, the time required to administer a shot will be far greater than
for cows housed in a stanchion barn.

Program costs are heavily dependent on the cutoff criteria specified by the
manager, as is highlighted through sensitivity analysis. Once the reproductive program
cutoff criteria has been established, the tool uses the cumulative probability of pregnancy
at the cutoff point as the probability that the cow is pregnant at cutoff to calculate the
value of the program. The remaining probability, or (1 - PROBPreg), iS the probability
that the cow is not pregnant at the cutoff of her breeding program and is also the

probability that the open RPO value is received.

3.3 Results and discussion

To demonstrate the decision making support provided by the Spreadsheet-based
tool an example scenario was analyzed using the on—farm tool. On-farm costs and values
and the farm managers’ cutoff rule for stopping breeding cows will affect the value of the
reproductive programs as calculated through the use of the decision tool. It is crucial that
managers use their own farm-specific values in order to find the optimal program for
their operation at the given time. Optimal programs will change across farms as well as
for a given farm over time. If labor costs, reproductive program supplies costs, or
reproductive efficiency to a given program were to change for a farm operation, the
optimal program for that farm may change as well. Sensitivity analysis must also be
performed to determine the sensitivity of the program costs and program value to key

parameters.

51

Costs and sensitivity to cutojjr criteria

AS stated previously, the cutoff that the manager selects to determine when to stop
breeding the cow will affect the program cost. Program costs subject to various cutoff
criteria are presented in Table 30 for both the visual heat detection and Ovsynch
programs. For simplicity, the breeding period for both programs in Table 3c is held
constant at 30 days. For illustration, program costs only, without inclusion of RPO
values or other revenues, are assessed subject to different cutoff criteria for these
reproductive management programs. The scenario being analyzed included a 100 cow
dairy herd comparing visual heat detection to the Ovsynch ovulation synchronization
program. The farm used semen that costs $15.00 per straw and the labor used for AI is
on-farm labor that costs $11.00 per hour. Fifteen minutes of on—farm labor were required
per AI, meaning that on-farm labor costs per AI were $2.75. Including semen costs, the
total farm costs for a single AI were $17.75. The labor costs per hour for visual heat
detection and administering synchronization shots were $9.00 and $14.00 per hour,
respectively. The farm expected to be able to purchase PGFZa for $3.50 per shot and
GnRH for $4.50 per shot. The on-farm costs of maintaining an open cow an additional
day was $2.25 per day. The discount rate used to discount revenues and costs was an
annual rate of 9% (Wolf etal., 2002). Reproductive program parameters included that
the farm currently uses visual heat detection and achieves a 55% AI submission rate
when all 100 cows are observed 2 hours per day. The CR that the farm has achieved in
the past to visually detected estrus is 42%. When evaluating Ovsynch, the farm estimates
a CR of 36%, but enters that the AI submission rate will be 100% because all cows are

expected to be submitted for AI after they are synchronized.

52

Table 3c, in addition to presenting the program costs as calculated in the decision
tool, also presents the probability that a cow is pregnant after the cutoff. The probability
of pregnancy is crucial in determining the value of the program, as it weighs the RPO
values for pregnant and open cows by the likelihood they are realized. Note that the
program costs associated with visual heat detection, in this example, are generally higher
than those with Ovsynch. These higher program costs with visual heat detection may be
surprising, given the cost to administer the program per cycle was lower for visual heat
detection. Note that although the cost to administer a visual heat detection program is
lower per month than Ovsynch, the probability that the cow is pregnant after each
breeding is lower as well. Due to the smaller pregnancy rate achieved with the visual
heat detection program in this example the farm must pay for additional months of
maintenance on an open cow, thereby increasing the costs associated with the visual heat
detection program. There is clearly an inherent tradeoff between the program costs
incurred monthly to administer the program and the probability that a cow is pregnant at
the end of each month. Such tradeoffs must be assessed using farm specific costs to be
sure that the farm is recognizing all additional costs associated with having a cow remain

open on the farm for an additional month.

53

Table 3 c. Program cost sensitivity to cutoff rule in example scenario

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Cost of Probability of Probability
Cutoff rule program pregnancy open
85% Cumulative probability
Visual of pregnancy $708.98 86% 14%
heat 90% Cumulative probability
detection of pregnancy $879.71 92% 8%
85% Cumulative probability
of prewancy $484.62 89% 1 1%
90% Cumulative probability
Ovsynch of pregnancy $579.40 93% 7%
3 Al Services $270.85 53% 47%
Visual 4 Al Services $359.79 65% 35%
heat 5 Al Services $448.07 71% 29%
detection 6 Al Services $535.69 77% 23%
3 Al Services $292.94 74% 26%
4 AI Services $389.14 83% 17%
5 Al Services $484.62 89% 1 1%
Ovsynch 6 Al Services $579.40 93% 7%

 

Value of reproductive management programs

Focusing on using the cutoff criteria of 6 AI and the reproductive management
programs of visual heat detection and Ovsynch, the values for the breeding programs
were calculated. The inputs for the tool were as follows: CR of 30% to Ovsynch, CR of
38% to visual heat detection, AI submission rate of 100% to Ovsynch, AI submission rate
of 65% to visual heat detection (with 100 cow group), and a cost per A1 of $17.75. The
farm expected to be able to purchase PGan for $2.52 per shot and GnRH for $3.59 per
shot. The on-farm costs of maintaining an open cow an additional day was $2.00 per
day. The discount rate used to discount revenues and costs was an annual rate of 9%
(Wolf et al., 2002). Using these inputs, and the RPO values from DairyVIP© version 1.1
(De Vries, 2006), values of reproductive management programs for cattle of various
lactation numbers can be determined. Reproductive program values for Ovsynch

assuming that shots take 2.1 and 6.3 minutes are assessed. Further, efficiency of visual

54

heat detection is assessed by altering the time required to obtain the visual heat detection
rate from 2.15 to 2.6 hours per day. Table 3d shows the values of the reproductive
management programs for various labor costs for cattle in their first lactation, assuming a
26 day period for Ovsynch and a 21 day period for visual heat detection. Table 3e, for
comparison, shows the values of the reproductive management programs for various
labor costs for cattle in their first lactation, assuming a 33 day period for Ovsynch and a

21 day period for visual heat detection.

Table 3 (1. Reproductive management program values for first lactation cows (26

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

day period for Ovsynch)
Al Al
submission submission
rate 65% rate 65%
with visual with visual
Ovsynch with 30% Ovsynch with 30% heat heat

Labor CR CR detection, detection,

cost and 2.1 minutes and 8.3 minutes 2.15 labor 2.8 labor

per per shot (26 day per shot (26 day hours per hours per

hour period) period) day day

$8.00 $517.52 $510.08 $555.66 $550.79
$7.00 $516.90 $508.22 $551.79 $546.11
$8.00 $516.28 $506.36 $547.91 $541.43
$9.00 $515.66 $504.50 $544.04 $536.75
$10.00 $515.04 $502.64 $540.17 $532.07
$11.00 $514.42 $500.78 $536.30 $527.39
$12.00 $513.80 $498.92 $532.43 $522.70
$13.00 $513.18 $497.06 $528.56 $518.02
$14.00 $512.56 $495.20 $524.69 $513.34
$15.00 $511.94 $493.34 $520.81 $508.66
$16.00 $511.32 $491.48 $516.94 $503.98
$17.00 $510.70 $489.62 $513.07 $499.30
$18.00 $510.08 $487.76 $509.20 $494.61
$19.00 $509.46 $485.90 $505.33 $489.93
$20.00 $508.84 $484.04 $501.46 $485.25

 

 

 

 

 

 

55

 

Table 3 e. Reproductive management program values for first lactation cows (33
day period for Ovsynch)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Al Al
submission submission
rate 65% rate 65%
with visual with visual
Ovsynch with 30% Ovsynch with 30% heat heat
Labor CR CR detection, detection,
cost and 2.1 minutes and 6.3 minutes 2.15 labor 2.6 labor
per per shot (33 day per shot (33 day hours per hours per
hour period) period) day day
$6.00 $484.89 $477.48 $555.66 $550.79
$7.00 $484.27 $475.63 $551.79 $546.11
$8.00 $483.66 $473.78 $547.91 $541.43
$9.00 $483.04 $471 .93 $544.04 $536.75
$1 0.00 $482.42 $470.07 $540.17 $532.07
$11.00 $481.80 $468.22 $536.30 $527.39
$12.00 $481.19 $466.37 $532.43 $522.70
$13.00 $480.57 $464.52 $528.56 $518.02
$14.00 $479.95 $462.66 $524.69 $513.34
$15.00 $479.33 $460.81 $520.81 $508.66
$16.00 $478.72 $458.96 $516.94 $503.98
$17.00 $478.10 $457.11 $513.07 $499.30
$18.00 $477.48 $455.26 $509.20 $494.61
$19.00 $476.86 $453.40 $505.33 $489.93
$20.00 $476.25 $451.55 $501.46 $485.25

 

Tables 3f and 3g, respectively, show the values of reproductive management

programs for cattle in their third lactation with similar sensitivity analysis as presented

above.

56

 

Table 3 f. Reproductive management program values for third lactation cows (26

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

day period for Ovsynch)
Al Al
submission submission
rate 65% rate 65%
with visual with visual
Ovsynch with Ovsynch with heat heat
Labor 30% CR 30% CR detection, detection,
cost and 2.1 minutes and 6.3 minutes 2.15 labor 2.6 labor
per per shot (26 day per shot (26 day hours per hours per
hour period) period) day day
$6.00 $351.74 $344.30 $361.66 $356.80
$7.00 $351 .12 $342.43 $357.79 $352.12
$8.00 $350.50 $340.57 $353.92 $347.44
$9.00 $349.88 $338.71 $350.05 $342.76
$10.00 $349.26 $336.85 $346.18 $338.08
$11.00 $348.64 $334.99 $342.31 $333.39
$12.00 $348.02 $333.13 $338.44 $328.71
$13.00 $347.40 $331.27 $334.56 $324.03
$14.00 $346.78 $329.41 $330.69 $319.35
$15.00 $346.16 $327.55 $326.82 $314.67
$16.00 $345.54 $325.69 $322.95 $309.98
$17.00 $344.92 $323.83 $319.08 $305.30
$18.00 $344.30 $321.97 $315.21 $300.62
$19.00 $343.68 $320.11 $311.34 $295.94
$20.00 $343.05 $318.25 $307.46 $291.26

 

 

 

 

 

 

57

 

Table 3 g. Reproductive management program values for third lactation cows (33
day period for Ovsynch)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Al Al
submission submission
rate 65% rate 65%
with visual with visual
Ovsynch with Ovsynch with heat heat

Labor 30% CR 30% CR detection, detection,

cost and 2.1 minutes and 6.3 minutes 2.15 labor 2.6 labor

per per shot (33 day per shot (33 day hours per hours per

hour period) perig day day

$6.00 $320.25 $312.84 $361.66 $356.80
$7.00 $319.63 $310.99 $357.79 $352.12
$8.00 $319.01 $309.13 $353.92 $347.44
$9.00 $318.40 $307.28 $350.05 $342.76
$10.00 $317.78 $305.43 $346.18 $338.08
$1 1 .00 $317.16 $303.58 $342.31 $333.39
$12.00 $316.54 $301.73 $338.44 $328.71
$1 3.00 $315.93 $299.87 $334.56 $324.03
$14.00 $315.31 $298.02 $330.69 $319.35
$15.00 $314.69 $296.17 $326.82 $314.67
$16.00 $314.07 $294.32 $322.95 $309.98
$17.00 $313.46 $292.46 $319.08 $305.30
$18.00 $312.84 $290.61 $315.21 $300.62
$19.00 $312.22 $288.76 $311.34 $295.94
$20.00 $31 1.60 $286.91 $307.46 $291.26

 

 

Clearly, the value of the reproductive management program is affected by the
costs of administering the program, the lactation number (because it influences the value
of the cow through the RPO), and the reproductive performance achieved with a given
reproductive management program. By allowing farm-specific parameters to be entered
into the decision tool individual dairy farms can determine the value of the reproductive
management program to their operation, and thereby make more informed decisions

when selecting reproductive management programs.

3.4 Conclusion
Challenges within the dairy industry with getting cattle pregnant efficiently exist

due to decreasing CR, difficulty detecting cattle in heat, and adjustments to management

58

as herd sizes grow larger over time. In response to the challenges regarding reproduction
in the dairy industry, various programs and technologies have been developed to aid
producers in effectively and efficiently getting cattle bred. The tool described here aids
producers in deciding which of these programs is economically optimal, subject to farm-
specific costs and values. In addition, reproductive management programs are valued
according to manager specified cutoff criteria, and managers are able to compare
program values across various programs. Farm managers are able to evaluate visual heat
detection programs and synchronization programs, and can perform sensitivity analysis to
key parameters such as CR, AI submission rates, and on-farm costs. Through the use of
this decision tool managers are able to determine the economically efficient reproductive

management tool for their specific farm operation.

59

CHAPTER 4: A STOCHASTIC ECONOMIC ANALYSIS OF DAIRY CATTLE
ARTIFICAL INSEMINATION REPRODUCTIVE MANAGEMENT PROGRAMS
4.1 Introduction

Dairy herd reproductive performance is integrally linked to profitability of
commercial U.S. dairy operations. Reproductive success is critical to the financial
success of the farm because reproduction is necessary to generate the next lactation, and
the largest source of revenue for a U.S. commercial dairy herd is milk sales (Wolf, 1999).
Further, the direct link between reproductive performance and dairy herd management is
widely accepted and has long been recognized in the dairy industry (Britt, 1985). In
response to reproductive efficiency challenges felt industry-wide, reproductive
management programs have been developed and are now available to aid producers in
efficiently breeding their dairy cattle. Given the increasing number of reproductive
management programs available, economic assessments are necessary to determine the
economically efficient sets of programs for farm managers with given characteristics.

For many dairy farms it is the failure to detect estrus efficiently that limits
reproductive performance (N ebel and Jobst, 1998). In recent years, increased levels of
production coupled with increased herd sizes have influenced how dairy farms manage
dairy cattle reproduction (Pursley et al., 1997), spurring increased attention to the
development of reproductive programs to increase reproductive efficiency and
performance. Today, various reproductive technologies are available for use on
commercial dairy farms, including artificial insemination (AI), estrus and ovulation
synchronization programs, sex-sorted semen, pedometers, computer-based record

systems, and multiple estrus detection aides. Costs associated with the use of

60

reproductive management programs, and with achieving a given level of reproductive
performance, are expected to vary Significantly across farms due to varying on-farm
costs, facilities employed, previous levels of reproductive performance, management
ability, and other economic factors. Reproductive performance and success under
various reproductive management programs will also vary across dairy farms due to herd-
specific or farm-specific characteristics.

In recent years, reproductive management programs have focused on hormonal
synchronization of dairy cattle. Synchronization programs aim to decrease the expenses
associated with estrus detection by either shortening the time during which cattle must be
observed through ovulation synchronization or to eliminate the need for estrus detection
completely with timed artificial insemination (TAI). Timed ovulation allows for
breeding cows by appointment, thereby eliminating the need for heat detection
completely, and has similar pregnancy outcomes to those obtained through AI performed
after heat detection (Pursley et al., 1997). There are several systematic breeding
programs available for use on commercial dairy farms today. Benefits for the dairy farm
using systematic breeding programs include increased efficiency in the use of labor of
heat detection and AI, and the convenience of managing the herd under synchronization
(Nebel and Jobst, 1998). Farms may use more than one synchronization protocol and the
program(s) best suited to a particular operation will depend on farm-specific costs and
herd-characteristics. Farm managers are left to determine whether a hormonal
synchronization protocol or visual heat detection is the optimal choice for their farm, and

then to determine the specific program best suited to their operation.

61

In addition to varying outcomes achieved with reproductive programs and varying
on-farm costs, farm goals and farm manager characteristics may affect the reproductive
programs chosen. In particular, risks associated with the outcomes of reproductive
programs and the risk preferences of the farm manager will likely aid in determining the
optimal program for a given operation. Farm managers have heterogeneous risk
preferences and may be classified as risk averse, risk neutral, or risk loving. The
riskiness of a given program’s outcome (i.e., the certainty with which a given program
will yield a certain conception rate (CR) or heat detection rate (HDR)) will affect the
decision of a particular farm to adopt a given program.

Reproductive management programs used on dairy farms will be affected by
multiple farm-specific costs. With the abundance of reproductive management programs
available, a dairy producer is left to ask: which reproductive management program is the
economically optimal program for my dairy farm? Identification of suitable or optimal
programs for farms with a given set of general characteristics (i.e., manager risk attitudes,
farm size, on-farm costs, goals for reproductive performance) is necessary to improve
recommendations to dairy farm managers and to further the understanding of why farms
adopt the reproductive technologies that they do.

Several spreadsheet-based models (Groenendaal et al., 2004, Meadows et al.,
2005, and Olynk and Wolf, 2007) have been developed to assess reproductive
management decisions. Spreadsheet models range from general assessments of
reproductive performance, often on a limited number of dairy operations or within a
given geographical region, to in-depth investigations into specific reproductive

technologies. Groenendaal et al. (2004) used a spreadsheet-based model called OptiCow

62

to support economically optimal breeding and replacement decisions on dairy farms
through the use of the marginal net revenue technique. Meadows et al. (2005) used a
spreadsheet-based model to estimate the economic effect of changes in herd-level
reproductive performance in Ohio dairy herds. Olynk and Wolf (2007) developed a
spreadsheet-based model to compare the expected net present values of pure and mixed
sexed semen AI strategies and identified breakeven values for which sexed semen
strategies had higher expected net present values than conventional semen strategies. A
commonly recognized theme in such models, which allow some flexibility in
parameterization, are that costs associated with supplies for reproductive programs and
the time required to administer the programs vary widely with manager skill and
experience levels, as will reproductive efficiency gains associated with a particular
program, underscoring the need for sensitivity analyses.

More generally, several models have been developed to determine the costs of
reproductive inefficiency. Some examples of such models include Meadows et al. (2005)
who focused on varying reproductive performance in Ohio dairy herds, Tenhagen et al.
(2004) who used field trials to compare reproductive efficiency and economic
implications, and the DairyVIP model by De Vries (2006) which was intended to
estimate the value of a pregnancy. Such models allow calculations of the costs of
reproductive inefficiency. The models currently available do not explicitly consider the
risk associated with the outcomes of various programs or technologies, nor do they
incorporate the risk attitudes of the decision maker. Further, it should be noted that the
flexibility to simulate numerous reproductive programs is necessary to adequately

address the question of which program is optimal for a given farm operation. A model

63

which is capable of assessing reproductive program adoption questions while
incorporating the riskiness of alternative programs and risk attitudes of the manager is
necessary in order to recommend the efficient set of programs for farms with given sets
of characteristics.

Stochastic dominance is a powerful tool for identifying preferred sets of risky
options for a decision maker with a given set of characteristics. Stochastic dominance
analysis has been used in several areas of agricultural decision making, including
preferred milking parlor investment options (Thomas et al., 1997), pest management
(Zacharias and Grube, 1984), water-conserving irrigation strategies (Harris and Mapp,
1986), alfalfa management strategies on dairy farms (McGuckin, 1983), and tillage
practices (Klemme, 1985). Stochastic dominance is a risk efficiency criteria that has
clearly been employed in various applications of decision making in agriculture.

The main objective of this study was to determine the economically optimal
programs for dairy farmers with various characteristics (i.e., farm size, risk preferences of
managers, on-farm reproductive program costs). Assessing reproductive technology
decisions under risk (i.e., risks associated with the outcomes of a particular program,
namely HDR and CR) was a primary focus of this analysis, and stochastic dominance
was used to separate the sets of risky options and identify efficient sets for decision
makers given risk preferences. Sensitivity analysis of the results to changes in key on-
farm parameters was used to assess differences in efficient sets for farms with varying
cost structures. Given that dairy farmers have heterogeneous risk preferences, efficient

sets of programs were sought for broad risk preference groups of farm managers.

64

4.2 Materials and methods

The variability in response to a specific program and the attitude of the decision
maker towards that variability were assessed through stochastic dominance analysis. By
including the risk associated with the response to the program, as well as the willingness
to accept risk by the decision maker, the problem for the farm manager becomes
maximizing expected utility. It is hypothesized that the inclusion of the risk associated
with the reproductive performance resulting for a specific program or technology, and the
risk attitude of the decision maker adds an important component to existing models and
decision support tools. It is also hypothesized that farm-specific costs for labor and
program administration will aid in determining which programs or protocols are
stochastically dominant for farms with a given set of characteristics. Risk attitudes (i.e.,
whether the manager is risk averse, risk neutral, or risk loving) are expected to be
integrally important in determining which reproductive programs are adopted by

particular farm managers.

Producer behavior

The assumption implicitly underlying this analysis was that individual dairy
producers seek to maximize their utility. The assumption employed in the construction of
the stochastic budgets was that dairy producers are seeking to select the minimum cost
program available to them, subject to achieving a certain level of reproductive
performance. It should be noted that utility maximization under risk neutrality is
equivalent to profit maximization. Cost minimization is a necessary condition for profit

maximization (Mas-Collel et al., 1995) because an important implication of a dairy farm

65

selecting the profit-maximizing production plan is that there is no way to produce the
same level of output at a lower total input cost. For example, if producers are able to
achieve a 35% pregnancy rate (PR) through visual heat detection or through hormonal
synchronization, it is assumed that producers will choose the method which minimizes
the costs of attaining this level of performance. Underlying the choice of cost
minimizing methods is the assumption that the revenue is constant regardless of the
option chosen. For example, it is reasonable to assume that the milk revenue resulting
from a pregnancy would be the same regardless of what methods were employed to
achieve that pregnancy. Theoretically, a simplified version of this problem under the
assumption of free disposal of output could be represented as:
n
Min 2 wixl. subject to f (x) Z q ,
i = 1
where:
f(x) = Production function where x=(xl, X2, xn),

q = Specified level of output,

wi = Input cost of input i (input costs are assumed to be strictly positive), and

xi = Inputs.

The method which maximizes utility will depend on farm-specific values for
labor costs, heat detection or injection administration efficiency (referring to labor time
invested to achieve a certain outcome), and other farm-specific factors. In this way, the
PR can be thought of as an output, because although it in itself cannot be marketed or

sold, it does have a value to the farm operation. In addition, the PR, which depends

66

directly on the HDR and CR, will vary according to the distribution of the CR and HDR

in response to various reproductive management programs.

Stochastic budget development

A series of budgets were developed in Excel (Microsoft, Seattle, WA) that were
capable of simulating reproductive performance under numerous reproductive
management programs. An underlying assumption was that the total milk produced over
a lactation initiated through any reproductive program which results in a pregnancy is
independent of the reproductive program used to initiate that lactation. In other words, it
is assumed that a cow bred via one program will yield the same amount of milk revenue
as if that same cow were bred via a different program.

Costs for various reproductive management programs were calculated in the
aforementioned budgets. The budgets were used to determine the costs associated with
various reproductive management programs, subject to achieving a 90% cumulative
probability of pregnancy. By summing the costs subject to achieving a fixed cumulative
probability of pregnancy, it is possible to directly compare the costs across programs.
Costs for heat detection programs and synchronization programs were generally
calculated as follows:

Monthly Visual Heat Detection Program Cost per Cow =
[(TIME * OBS * 30.4) * LABOR HD]
+ AID + (C1 * HDR

 

COWS Visual ) ,
where,

TIME = Minutes per day invested in heat detection for a single group of cows,

OBS = Number of times the group is observed per day,

67

LABORHD = Cost of labor (in dollars per minute) to perform heat detection,

COWS = Number of cows in the group,
AID = Cost per cow per period of heat detection aid utilized,

CI = Cost of an AI, and
HDRVisual = Heat detection rate to visual heat detection.

Monthly Cost of Synchronization Program using GnRH and PGFZa per Cow =

(P X X

1|:
GnRH P P

PGFZa ) +

:1:
GnRH) +( GF2a

(MIN * SHOTS * LABOR I ) + (C1 * HDR

Total Synch )

where,

PGnRH = Price per injection of GnRH,
XGnRH = Number of GnRH shots given in the synchronization protocol,
PpGF20L = Price per injection of PGFZa,

XpGF2a = Number of PGF 201 shots given in the synchronization protocol,
MIN = Minutes to give a single injection,
SHOTSTotaI = Total number of injections in the series,

LABOR1= Cost of labor (in dollars per minute) to give injections,

CI = Cost of an AI, and

HDRSynch = Heat detection rate under synchronization.

Using the heat detection and synchronization costs per cow per month, the total

reproductive program costs were calculated by summing the program costs per month

68

over the number of months necessary to obtain a 90% cumulative probability of
pregnancy. For example, if, based on the inputted CR and HDR, it is calculated to take
four months in order to achieve a 90% cumulative probability of pregnancy, the total
program cost was the monthly program cost summed over four months.

Using the above described budgets it was possible to evaluate decisions regarding
which reproductive management systems are best suited for various operations.
Additionally, since the budgets allow entry of cow numbers per group rather than
assessing all cows on the farm at once, farm managers can determine the optimal
program for different groups of cows on their operation (i.e., select different programs for
cows which have had better prior reproductive performance than others or select different
programs for cows versus heifers).

Stochastic variables were incorporated into the budgets to account for the
riskiness of the HDR and CR in response to various reproductive management programs.
Simulation models that do not account for the riskiness of outcomes give a deterministic
result in response to the input variables. Stochastic models are used to account for
variables for which the values are not known with certainty, although for which a known
probability distribution exists. Through the use of stochastic variables, the interactions of
risky variables (in particular the interaction of the CR and HDR to achieve a given PR)
allow the user of the tool to determine how a program may perform under alternative
outcomes. Stochastic dominance is a popular method for ranking risky alternatives

without in-depth knowledge about the preferences of the decision maker (McCarl, 1990).

69

Stochastic dominance analysis

Stochastic dominance is a powerful risk efficiency criteria. Such criteria separate
a set of possible actions into an efficient set which is preferred by the decision maker and
a risk-inefficient set which is undesirable to the decision maker (Thomas, et al., 1997).
Efficient sets are defined to include the choice of every decision maker (farm manager) to
whom the specified decision rules apply. Efficient sets are identified by comparing the
cumulative distribution function (CDF) of possible options for the decision maker which
have risky outcomes. First degree stochastic dominant (F SD) and second degree
stochastic dominant (SSD) efficient sets were both evaluated in this analysis. An
underlying assumption for both of FSD and SSD is that the alternatives compared, in this
case the reproductive management programs compared, are mutually exclusive, meaning
that one program or the other must be chosen and that a program which combines the two
options is not feasible.

F SD has the intuitive meaning, according to Mas-Collel et al. (1995) that “the
distribution F (.) yields unambiguously higher returns than the distribution G(.).”. The

distribution F(.) first-order stochastically dominates G(.) if, for every non-decreasing

function u: Iu(x)dF (x) 2 Iu(x)dG(x) , where u indicates the utility function of the

decision maker. Therefore, F SD can identify the efficient set for all those decision
makers who prefer ‘more to less’ or, in this case, prefer the lower cost program. The
SSD analysis then seeks to introduce a comparison based on relative riskiness or
dispersion. SSD identifies the efficient set for decision makers who prefer ‘more to less’
at a diminishing rate, and are therefore risk averse. For any two distributions F (.) and

G(.), F(.) second-order stochastically dominates G(.) if, for every non-decreasing concave

7O

function u: Iu(x)dF (x) _>. Iu(x)dG(x) , where u indicates the utility function of the

decision maker. Second degree stochastic dominance analysis does not require identical
means, although the assumption of identical means reduces the analysis to assessing one
distribution as a mean preserving spread of the other. A risk averse individual is known
to prefer a mean preserving contraction over a mean preserving spread. Through the use
of these rules, efficient sets were identified for those decision makers fulfilling the
aforementioned assumptions.

An economic analysis was conducted to determine the optimal reproductive
program for farms with a given set of characteristics. Stochastic budgeting was used to
account for the uncertainties in the decision, namely the HDR and CR resulting from a
reproductive program, and to give an indication of the distribution of the outcomes
expected. This analysis began by running the decision tool for multiple iterations with
key parameters (i.e., CR, HDR, and the resulting PR) distributed across the range of
values identified in previous studies. In this way, the data from previous reproductive
performance studies was used in conjunction with survey data to parameterize the
economic analysis. The stochastically efficient or efficient set was then identified by
comparing the CDF of the risky alternatives, i.e., the various reproductive programs.
Through the use of @RISK (Palisade Corp., Newfield, NY) the budgets were evaluated
for 10,000 iterations. The resulting CDF of the risky alternatives were graphed for ease
of visual observation in Stata software (Intercooled Stata for Windows, version 8.2, Stata
Corporation, College Station, TX).

Key outcome parameters from the reproductive management programs were made

stochastic by parameterization of triangular distributions for those variables. In

71

particular, farm-size differences in program costs were assessed to highlight sensitivity to
farm-specific costs. Triangular distributions were used due to the limited sample data
available. Often triangular distributions are used in such cases as limited sample data
because only a minimum, maximum, and most likely value are necessary to parameterize
the model. Multiple scenarios are necessary in order to identify the efficient sets for
decision makers with various baseline farm characteristics. A general base-case example
scenario was constructed for illustration from previous survey data under the assumptions
that on-farm labor costs $12.78 per hour for either heat detection or to give injections,
costs per AI were $17.30, labor hours devoted to heat detection were 2.15 per day if
visual heat detection was used, labor time to give a single injection was 2.10 minutes if
synchronization was used, GnRH cost $3.59 per shot and PGFZa cost $2.52 per shot. The
number of cows per group, or number able to be observed at a single time was assumed
to be 100.

Visual heat detection without aides, Ovsynch, and Cosynch were investigated
under this general scenario. The protocols for OVSynch and Cosynch are nearly identical,
varying only by timing of A1. The protocols for both Ovsynch and Cosynch are
diagramed for reference in Figure 4a. Cosynch is essentially a specific modification of
Ovsynch in which cows receive TAI concurrently with the second GnRH injection.
Cosynch has the advantage of allowing dairy farms to restrain cows one less time than
the Ovsynch program, and allows for all cow-handlings to occur at the same time daily
(F ricke, 2003). Although there may be advantages for the Cosynch program from a cattle
handling standpoint, optimal conception rates are not achieved using Cosynch (Pursley et

al., 1998). Due to the similarities between the Cosynch and Ovsynch programs, they

72

offer an interesting comparison, as dairy farms considering one of these programs would

likely consider the other as well.

Figure 4 a. Ovulation synchronization programs used as examples

Ovsynch (Does not include PGFZa injections before the first GnRH injection)

 

 

PGFZa
GnRH GnRH
j V j Timed Al
I 7 days 2 days W) — 24 hours '

Cosynch (Specificform of Ovsynch in which breeding occurs concurrently with the

second injection of GnRH. )

Timed Al concurrent with

 

GnRH shot
PGFZa l
Gan-l GnRH
I | g
I 7 days 2 days I

 

The distribution used for the CR to visual heat detection was parameterized using

survey data and a triangular distribution with a minimum of 26%, maximum of 60%, and

most likely value of 41%. For Ovsynch, the CR was also distributed triangularly,

although with a minimum of 23%, maximum of 57%, and a most likely value of 38%.

The minimum and maximum values were adjusted slightly from the CR distribution used

for visual heat detection, as mixed opinions exist regarding the differences in CR between

A1 to visually detected heat versus TAI. The most likely value for Ovsynch was taken

73

from Pursley et al. (1997). The triangular distribution for CR under Cosynch reflected
the lower CR expected with Cosynch versus Ovsynch, which is expected to be
approximately 6 percentage points, on average (J. R. Pursley, personal communication,
2007). The triangular distribution included a minimum CR of 21%, a maximum of 5 5%,
and a most likely value of 32%.

The triangular distribution used for visual heat detection without aides was
parameterized using survey data and had a minimum of 20%, maximum of 75%, and a
most likely of 52%. The HDR used for Ovsynch was a triangular distribution of 96%,
99%, and 100% for the minimum, most likely, and maximum values, because most cows
are being bred to TAI; similarly values for Cosynch were 97%, 99%, and 100% because
cows are almost certainly bred TAI, except under unexpected circumstances. The PR
resulting from each of these programs was affected by the distributions of the HDR and
CR under each program.

Ntunerous base scenarios are possible to allow stochastic dominance assessments
for farms of various herd-sizes. Multiple iterations must be run for farms with a given
herd-Size and on-farm labor cost. By holding these factors describing the general
characteristics of the farm fixed and running several iterations with the HDR and CR
being treated as stochastic, CDFs can be generated and compared to determine efficient
sets for decision makers with different risk preferences and given certain farm
characteristics.

In order to assess the sensitivity of the analysis to farm-specific costs, sensitivity
analysis was conducted using the same reproductive programs as above. In particular,

sensitivity to on-farm costs and cow group sizes were assessed. Farms with less than 100

74

cows had higher costs for purchasing hormones and higher per AI costs according to
survey data. For comparison, a scenario indicative of small-farm costs and values was
constructed from survey data. The survey data used, broken down by farm size, is
presented in Table 4a. Parameterization of the small farm comparison used the following
parameters: labor costs were maintained at $12.78 per hour, per AI costs were $21.00 for
visual heat detection and Cosynch, per AI costs were $24.20 for Ovsynch and were
adjusted in the same way as prior, 2.67 labor hours were devoted to heat detection per
day, labor time to give a single injection was 1.7 minutes, $4.49 per shot of GnRH and
$3.10 per shot of PGFZa. The number of cows to be observed at a single time was

assumed to be 35, in order to be in keeping with the smaller total farm size.

Table 4 a. On-Farm values affecting costs by farm size

 

 

 

 

HD time
(minutes Cost of Cost of
Farm size per day for Cost Minutes GnRH PGFzal
(cows) group) per AI per shot per shot per shot
<100 160 21 1.7 4.49 3.1
100-200 122 17 6 3.13 2.13
>200 94 11 1.9 2.7 2.1

 

 

 

 

 

 

 

 

Source: Survey described in Chapter 2

Farms will vary on whether they incur additional costs associated for breeding
cows on Ovsynch due to various factors. Farms may have facilities that enable automatic
sorting, thereby making an additional cow-handling very efficient. Other farms,
however, may incur significant costs associated with sorting and handling cattle an
additional time for breeding with Ovsynch versus Cosynch. Such differences warrant
sensitivity analysis to such factors. Further, a comparison case was constructed for
sensitivity analysis where additional costs were incurred for breeding with Ovsynch

versus the other programs. Such sensitivity analysis highlights the differences that the

75

initial farm costs and group size assumptions can have on the analysis. There is an
inherent tradeoff between these programs in that cattle must be handled an additional
time with Ovsynch for breeding, but the expected CR is slightly higher than that with
Cosynch. In order to account for the additional handling associated with breeding in the
Ovsynch program, the cost per Al was adjusted from the $17.30 described above to
include 15 additional minutes of labor cost, which was assessed at the $12.78 per hour as
described above. Therefore, the cost per Al for the Ovsynch protocol was assessed at
$20.50, to include the $17.30 per AI from the base scenario plus $3.20 in added labor
costs for an additional cattle handling. All other costs were held constant with the base

example scenario described above.

4.3 Results and discussion

Farm-specific technology adoption decisions were found to be highly sensitive to
on-farm costs and individual farm characteristics. Further, risk attitudes of the farm
decision maker affected the efficient set of reproductive programs for a given scenario.
Examining the uncertainty of the response in reproductive performance to reproductive
management programs, and therefore the changes in costs associated with reproductive
management programs, will help determine why different farms select different
reproductive management programs when seeking to maximize total farm profitability
through reproductive management. In this way risk preferences are expected to affect
what technologies or programs are in the efficient sets.

Table 4b illustrates the summary statistics for the costs of the programs resulting

from the 10,000 iterations run on the afore-described base example analysis. Directly

76

following the summary statistics is Figure 4b, which illustrates mean-variance analysis,
where the program with the lowest average cost and the smallest variance (standard
deviation is used here) is selected. In this analysis, Ovsynch was found to have the
lowest average cost for a program, and also to have the lowest standard deviation of the

programs considered.

Table 4 b. Summary statistics for program costs in example analysis

 

 

 

 

 

 

 

 

 

 

 

 

 

Visual
heat
detection
without
aides Ovsynch Cosynch
Mean -164.13 Mean -123.97 Mean -140.48
Median -162.74 Median -112.68 Median -140.46
Mode -205.40 Mode -167.98 Mode -224.49
Standard Standard Standard
deviation 36.08 deviation 33.37 deviation 38.92
Most Most Most
costly -262.31 costly -226.44 costly -254.79
Least Least Least
costly -76.96 costly -55.58 cosgy -56.27

 

 

77

Figure 4 b. Mean-variance analysis for example

 

 

O
(‘11 ..
G Ovsynch
O
53 .4
O
C 3 -1 O
E ' Cosynch
Direction of
o optimization
’2" O
O
(‘2 ..
I . Visual heat detection
F I l I I
32 34 36 38 40

StDev

Figure 4c illustrates the CDFs of the various reproductive management programs
available. In Figure 4c, comparing visual heat detection without aides to Ovsynch,
Ovsynch has F SD and SSD over the visual heat detection program. When comparing
visual heat detection with no aides to the Cosynch program, there is no FSD and no SSD.
It is important to note that while no FSD or SSD dominance exists between visual heat
detection without aides and Cosynch, this is due to the slightly higher chance of having a
very high cost program, and that the vast majority of the time the synchronization
programs are cheaper than visual heat detection with no aides. When comparing
Ovsynch and Cosynch to one another, there is F SD of and SSD of Ovsynch over

Cosynch, although the FSD looks deceiving graphically. Note that all FSD and SSD

78

outcomes described above are subject to the parameters outlined in this example. For this
example case, the efficient set for FSD includes Ovsynch because it F SD visual heat

detection and Cosynch. Additionally, Ovsynch is in the efficient set for SSD because it

SSD visual heat detection and Cosynch.

Figure 4 c. Cumulative density functions for example analysis

 

 

 

 

 

 

 

Q ..
£0. —
B
8
o
i V. -i V n.-
N ..
o .4 ‘1 “--J(--
-250 —200 -150 —100 -50
Cost of Program ($)
VisualHD - Ovsynch
Cosynch

 

 

 

Farm location and farm size have effects on the on-farm costs associated with
reproductive management programs. In order to assess sensitivity to on-farm costs, FSD
and SSD was assessed for small farms, as indicated in the Methods sections, and results
are shown in Figure 4d. Sensitivity to on-farm costs can be seen in the higher costs
associated with reproductive programs on small farms. Perhaps most significant is the

extreme difference in the visual heat detection without aides program. Costs are

79

considerably higher for small farms due to watching fewer cattle at a time for visual heat
detection and higher per injection hormone costs for synchronization. In addition, per AI
costs were higher on small farms. For small farms (less than 100 cows) the efficient set
for FSD is Ovsynch because Ovsynch F SD both visual heat detection and Cosynch.

Since FSD is present, Ovsynch SSD Cosynch and visual heat detection, which was also

true in the all-farm analysis presented prior.

Figure 4 11. Small farm sensitivity analysis

 

 

    

Q—r
.....1.
r.
50.4
E
N
8 1"";
Earl
. 1,
r-
61.4
I
.1,

 

 

 

 

 

 

 

O
l I T l
-600 0 -200 0
Cost of Program ($)
VisualHD Ovsynch
Cosynch

 

 

 

Figure 4e displays the CDF comparison when Ovsynch incurs additional breeding
costs compared to the other two programs. Notice that both Ovsynch and Cosynch have
F SD over visual heat detection in this case; given this FSD, Ovsynch and Cosynch are

both SSD over visual heat detection. The interesting case in this example is the

80

comparison between Ovsynch and Cosynch. Notice there is no F SD between Ovsynch
and Cosynch in this case. When assessing SSD, Cosynch is SSD over Ovsynch. The
additional 15 minutes of labor costs incurred per AI in the Ovsynch program in this case
has led to Cosynch being SSD over Ovsynch. SSD of Cosynch over Ovsynch would
indicate that risk averse managers prefer the Cosynch program to the Ovsynch program,
in which they have to incur additional costs associated with handling cows an additional
time, an assumed 15 minutes in this example, for breeding purposes. Given this analysis,
in the case in which Ovsynch incurs 15 minutes of additional labor costs over the other
two programs, the efficient set for FSD is Ovsynch and Cosynch because both dominate
visual heat detection. The efficient set for SSD is Cosynch because it dominates visual

heat detection and the Ovsynch program.

Figure 4 e. Sensitivity analysis: Ovsynch incurring additional breeding cost

 

 

 

 

 

 

 

 

 

Q ._
am. 4
E
(U
.0
0
1: V. —
(V. _.
o _.
I l l l l
-250 -200 - 0 -100 ~50
Cost of Program ($)
VisualHD Ovsynch
Cosynch

 

 

 

81

Through FSD and SSD analysis, the efficient sets of reproductive programs have
been identified for those producers who, I) prefer a less costly program (FSD), and 2)
prefer a less costly program and are risk averse. It is important to note that the programs
which are present in the efficient set depend integrally on the parameters set up in the
example problem. Those programs in the efficient set will be different for farms with
varying on-farm labor costs, labor efficiencies, and decision maker preferences. This fact
strengthens the notion that regional data or farm size specific data sets would be useful in

parameterization of this analysis to allow costs which are more Specific to a given locale.

4.4 Conclusion

Dairy herd reproductive performance is closely tied to whole-farm profitability
for commercial U.S. dairy operations. Identification of optimal programs for farms with
a given set of general characteristics (i.e., operator risk attitudes, farm size, and on-farm
costs) is necessary in order to enable recommendations to dairy farm operators and to
further the understanding of why farms adopt the reproductive technologies that they do.
In this analysis, stochastic variables were incorporated into a series of budgets to account
for the riskiness of the reproductive program outcomes, namely HDR and CR. FSD and
SSD were used to determine the efficient sets of reproductive programs for decision
makers with heterogeneous risk preferences.

Perhaps the most interesting and surprising finding, highlighted through the base
example case, was the FSD and SSD of Ovsynch over Cosynch, which indicates that

decision makers of all risk preferences prefer Ovsynch rather than Cosynch. This

82

dominance of Ovsynch indicates that decision makers do not want to take the CR risks
associated with Cosynch. Although the synchronization programs employ the same
number and type of hormonal injections, the timing of these injections affects the CR.
When Ovsynch cost 15 minutes more in labor time for each AI, however, Cosynch
became SSD over Ovsynch indicating that risk averse managers would then prefer
Cosynch. This analysis has highlighted that risk aversion is affecting which programs
remain in the efficient set, and since dairy farmers are likely risk averse, the SSD analysis
is particularly important for the dairy industry. These types of decisions among
synchronization programs are one of the key contributions of this model.

Given the flexibility of the on-farm decision tool, parameterization of the model is
possible for regional or even farm-specific data. When assessing only small farms, all
program costs were found to be higher than the general assessment, indicating that
regions with large proportions of small farms will find such analysis particularly
important in assessing the optimal program for their operations. Further, farm costs, such
as labor costs associated with breeding for a particular program are important in
identifying efficient sets for farms with given characteristics.

Overall, the incorporation of the risk preferences of the decision maker is an
important contribution to farm-level decision making. By identifying efficient sets of
programs for decision makers with various risk preferences we are better able to make

recommendations for managers with given farm characteristics.

83

CHAPTER 5: SUMMARY AND DISCUSSION

Reproductive management of dairy cattle is crucial to whole-farm profitability as
it enables milk sales, provides replacement animals, and is an important factor in
potentially costly culling decisions. Further, reproductive management has become a
challenge industry-wide as dairy producers face: (i) conception rates in cows that have
decreased from 60% to 40% over the years in which AI has been practiced in the United
States (N ebel, 2002); and (ii) increasing challenges with detecting cattle in estrus as herds
become larger. The dairy industry has responded with innovative technologies and
reproductive management programs that enable producers to synchronize ovulation,
thereby eliminating the need for heat detection. Beyond synchronization programs, heat
detection aides enable more efficient and accurate visual heat detection; automated
computer-based record keeping systems make in-depth record keeping on individual
cows possible, and technologies such as ultrasound and embryo transfer are offering
options to dairymen that never existed before.

With all of the recent innovation and the multitude of programs and technologies
available, dairy producers must decide which programs are economically optimal for
their farm operations. The economically optimal choice for a given farm operation will
be dependent on several factors, including on-farm costs and values, farm manager ability
and knowledge, facilities used to handle and manage cattle, and the goals of the farm
operation. In addition, the risk preferences of the manager, meaning whether the
manager is risk loving, risk neutral, or risk averse, will affect the amount of risk that a

manager will accept in the outcome of a program. Stochastic dominance analyses of

84

reproductive management programs highlighted that risk preferences of decision makers,
in addition to the on-farm costs and values used to parameterize the problem, affected the
reproductive management programs that were within the efficient set for a farm manager.
In the base case example shown in this analysis, Ovsynch dominated Cosynch and visual
heat detection in the first degree, making it the selected option for decision makers of all
risk preferences who simply prefer ‘more to less’ of the outcome. In the case of
reproductive programs, all decision makers who prefer a higher pregnancy rate to a lower
pregnancy rate would choose Ovsynch, given the parameters used to characterize the
farm situation in the model. When the base case example was altered to include a charge
for the additional handling required for breeding cattle under the Ovsynch program,
Cosynch dominated Ovsynch in the second degree, indicating that risk averse decision
makers would prefer Cosynch over Ovsynch. These differences in which programs are
dominant, based on the parameters used to describe the farm situation, highlight the need
to perform farm-specific analysis. Further, risk preferences clearly aid in explaining farm
managers’ choice of reproductive management programs.

The analyses presented built upon prior reproductive management studies and
sought to inform economically sound decision making regarding reproductive program
and technology adoption. The varying costs and revenues associated with reproductive
performance across farms illustrated the need for farm-specific analyses regarding
selection of economically optimal reproductive management programs. In particular,
sensitivity to on-farm labor costs highlights the necessity to evaluate adoption decisions
for individual farms; the reproductive management program that is Optimal for one farm

is likely not also optimal for a farm with differing labor costs. Different reproductive

85

management programs will vary in sensitivity to labor costs; synchronization programs
which require labor for administering shots are generally less sensitive to on-farm labor
costs than visual heat detection based AI programs. The differences in sensitivity to
labor are dependent on the amount of time required to administer each program, namely
the amount of time needed to administer a series of shots versus the amount of time
necessary to perform heat detection for a group of cows each day.

Previous work by Meadows et al. (2004) highlighted that the marginal benefits of
improved reproductive performance are decreasing as reproductive performance
improves. Current farm-level reproductive performance was found to be important
through these analyses in assessing reproductive management programs for a given farm
Operation. Farms that had obtained high levels of reproductive efficiency through visual
heat detection, for example, had less incentive to adopt a synchronization program than
those farms with less efficient visual heat detection. This highlighting of previous
reproductive performance when selecting reproductive management programs is
illustrative of the multitude of programs that are seen on farms today. Farms that have
experienced success with visual heat detection, for example, will have less incentive to
adopt a different reproductive management program. Combine the riskiness associated
with the outcome of reproductive management programs with the low levels of incentives
to adopt a new program, and it can be better understood why there is such a range of
breeding technologies and programs used in the industry today.

With survey responses indicating that 77% of farms had made a change in their
reproductive management system between 2000 and 2005, there is clearly an ongoing

need for continued research in the area of reproductive management. Several farms

86

reported changes involving the initiation of a synchronization program in place of visual
heat detection or moving from breeding with natural service to using AI. While many
producers indicated adoption of technology as their most recent reproductive
management change, other producers indicated a departure from the use of technologies
such as synchronization or Al. With the array of adoption and disadoption decisions
being made, it is clear that the program that is optimal for one farm is not necessarily
optimal for another. Even for an individual farm, the program that is optimal currently
may not be optimal in the future if on-farm costs change. Further, producers will benefit
from decision support tools which aid in reproductive management program and
technology adoption decisions as they seek to identify the economically optimal
programs for their operations.

Overall, the reproductive management programs employed differ across farms
due to varying on-farm costs and values, farm goals, management preferences, facilities
used, and previous reproductive performance and experience. Differing characteristics
across farms aid in explaining why the reproductive program that is optimal for one farm
may not be economically feasible for another. Through the use of surveys, sensitivity
analyses to reproductive program costs, and assessment of farm manager decision making
under risk, the reproductive program decisions made on farms are better understood.
Many factors are taken into account when farm managers make decisions regarding
which reproductive management programs and technologies to use on their operation.
By better understanding the factors that farm managers consider important and
incorporating them into decision support tools for use on individual farm operations, the

industry is better able to serve producers.

87

APPENDICES

88

APPENDIX 1: DAIRY PRODUCER SURVEY

Survey - Reproduction and Heifer Rearing on Dairy Farms

 

[ General Farm Characteristics — Part A

 

A1. How many head of dairy stock were on hand January 1”, 2005?
Total Milk Cows (including first calf heifers and dry cows)
Total heifer calves and replacement heifers
Bulls
Dairy steers and bull calves

A2. Types of facilities for cows and heifers. (Please mark predominant type with
“P” and all others that apply with an X)
Cows Heifers

 

 

 

Stanchion/tie stall barn [:1 El
Free stall barn E] El
Bedded pack barn E] El
Dry lot [:I [:I
Pasture [:1 Cl
Other (Please Specify) D [:1
Age of current housing facilities (years)
A3. Total pounds of milk sold by this farm in 2005 pounds

 

A4. Family and hired labor usage in 2005
Number of Avg. Months Avg. Hours
Workers Worked/ Worked]
a. Unpaid labor Worker Month
Spouses
Children over 12
Other unpaid labor
b. Paid labor
Hired manager/operators
F uIl-time
Part-time (year around)
Seasonal workers

89

A5. What were the wage/salary levels for workers of all levels of your operation

 

in 2005? Wage Rate/hr or Salary level/year
Hired Managers/operators Ihr or lyear
Full-time workers Ihr or lyear
Part-time workers (year around) Ihr or lyear
Season workers Ihr or lyear
Other (Please specify) Ihr or lyear

 

A6. Who is responsible for record keeping in your operation?
Job Title
Other Responsibilities

 

 

A7. What is the primary herd management record keeping system utilized in
your operation?

[3 Paper
El Dairy Comp 305 or SCOUT
E] DHI

E] PCDART

D Other (Please Specify)

 

A8. Who comprises the management team (decision making team) on your
operation?
(Please check all that apply and give a brief description of their primary
role/responsibility in affecting decision making.)
Roles/Primary Responsibilities
1:] Owners/operators
[J Veterinarian
CI Nutrition Consultant
El Banker
1:] Accountant
El Al Sales Representative
[3 Herd Manager/Herdsman
C] Other Employees
(Please Specify)

 

 

 

 

 

 

 

 

A9. What was your cull rate for 2005?

 

A10. Were the reasons for culling recorded in 2005?
[I Yes (If yes, please proceed to question A11)
1:] No (If no, please skip to question A 12)

9O

A11. If reasons for culling were recorded, what percentage of culls were due to
the following reasons in 2005?

Sold for dairy purposes Injury
Low milk production Death
Feet and legs Mastitis
Reproductive performance Disease

Udder problems

A12. What criteria are utilized for voluntary culling decisions? (Please check all
that apply.)

El Current heifer and/or cow prices

1:] Number of springing heifers in cattle inventory
1:] Space available

[:| Other (Please Specify)

 

 

[ Calves and Heifers - Part B j

 

B1. Did this farm utilize a custom heifer raiser in 2005?

E] No (If no, please skip to question 85)
[:I Yes (If yes, please proceed to question 82)

32. Please indicate your reasons for utilizing a custom heifer raiser.
(Please check all that apply.)
Management time constraints
Cl Lack of adequate facilities on home farm
El Manure management concerns
[I Better growth/performance from custom raiser
E] Expansion of milking herd/cow numbers
D Other (Please Specify)

 

B3. If you have previously raised calves/heifers in your operation and have
switched to utilizing a custom heifer grower -) Have you noticed better
performance and growth with the utilization of the custom raiser?

EINO

[:I Yes 9 Please comment on any differences you have noticed.

 

91

B4. Please rate your overall satisfaction with the utilization of a custom heifer
raiser on the following scale of 1 — 6, with one being extremely dissatisfied and
six being extremely satisfied.

Extremely Extremely
Dissatisfied Satisfied

 

D1 D2 D3 D4 D5 D6>

B5. Did you utilize an accelerated heifer growth program in 2005 at any stage of
heifer growth?

D Yes (If yes, please proceed to question 86 and skip question 87)

[:1 No (If no, please skip to question B7)

B6. If you are utilizing an accelerated growth heifer program please indicate the
stages of growth being accelerated below.

 

B7. If you are not utilizing an accelerated heifer growth program please indicate
your reasons why. (Please check all that apply.)

1:] Expense
[:1 Lack of knowledge/information on management of a program
[3 Lack of management time to oversee the program

E] Not convinced of the benefits

1:] Other (Please Specify)

 

BB. What are pre-weaned calves being fed?
1:] Milk replacer % fat % protein
1:] Non—pasteurized waste milk
[:1 Pasteurized waste milk

B9. What criteria are utilized in weaning calves?

Criteria Used
1:] Age weeks old
[:1 Daily grain intakes lbs/day

C] Other (Please Specify)

 

B10. What is the average age at weaning on your farm?

 

811. What is the average weight at weaning on your farm?

 

92

B12. Are calves/heifers being weight taped regularly with height and weight
recorded?

Cl Yes (If yes, please proceed to question 813)
E] No (If no, please skip to question B14)

B13. If calves/heifers are being weight taped, how often is this recorded at each
stage of life? (Please mark any stages at which weight taping occurs.)

Taged During Stage Frequency of Weight Taging

[:1 Pre-weaning
|:l Post-weaning — breeding age
[I Bred - springing

B14. What proportion of heifer calves born survived to first service in 2005?

 

 

I Reproduction — Part C

 

C1. What is your average age and weight of heifers at their first
insemination/breeding?
Age Weight

 

C2. What is your average age and weight at first calving?
Age Weight

 

C3. What percentage of lactating cows were open at greater than 150 days in
milk (OPEN>150) in 2005?

 

 

C4. What is your voluntary waiting period for lactating cows?

C5. What is the average number of days to first service for lactating
cows?

 

06. What is your calving interval?

 

C7. What is the average length of your dry period?

 

CB. What is your average number of days open?

 

93

09. What are the heifer breeding criteria used on your farm? (Please check all
that apply.)

 

Criteria Used
1:] Age months
[I Percentage of mature bodyweight % at breeding
% at calving

[:I Frame size inches at withers

C] Other (Please Specify)

 

C10. Do you utilize artificial insemination for breeding cows and/or heifers?

Cows Heifers
1:] Yes I: Yes
[I No I] No

Please Note: If you answered no to both heifers and cows in question C10,
please skip ahead to question C17.

C11. Percentage of breeding using artificial insemination (Al): Percentage
Cows
Heifers

C12. Who is responsible for Al on your operation for cows and/or heifers?
(Please check all that apply.)

Number of
Breeders
[:1 Owner/operator
1:] Herdsman
E] Heifer manager
1:] Breeding manager
E] Outside Al technician (Genex, Alta, Select Sires, etc)
1:] Other (Please Specify)

 

 

 

 

 

 

 

C13. Was sexed semen being used in your operation in 2005?

EINO

1:] Yes -) Please specify which groups of animals it was used on.

 

94

014. What is the average price per straw of semen used on your farm to breed
cows and heifers?

Cows $ lstraw

Heifers $ lstraw

C15. Please state your top 3 criteria used in sire selection for cows.

 

 

 

C16. Please state your top 3 criteria used in sire selection for heifers.

181
2nd
3rd

 

 

 

C17. If you do not use 100% Al, for what reason(s) do you use natural service?
(Please check all that apply.)
Cows Heifers

 

 

Cost of semen D [I
Lack handling facilities [:1 E]
Lack labor for estrus detection and servicing [I [:l
Bred 1"t service Al, then introduce clean-up bulls [:1 El
Other (specify) Cl C]

 

95

 

[ Heat Detection Methods - Part D
D1. Which heat detection methods are currently being utilized on your farm.
(Please check all that apply, and provide percentage of animals in each
category.)
Cows Heifers

[:I Visual heat detection without aides Cl Visual heat detection without aides

 

 

 

 

 

% cows % heifers
1:] Passive mount detectors 1:] Passive mount detectors
% cows % heifers
[:l Kamar [:l Kamar
[:1 Chin ball markers 1:] Chin ball markers
E] Tail chalking/crayon [:1 Tail chalking/crayon
[3 Other (Please Specify) D Other (Please Specify)
1:] Electronic aides [:1 Electronic aides
% cows % heifers
El Heat Watch [:1 Heat Watch
[:1 Pedometers [:1 Pedometers
CI Afi System [I Afi system
C] Other (Please Specify) C] Other (Please Specify)
C] Other (Please Specify) I] Other (Please Specify)

 

 

DZ. If visual heat detection without aides is being utilized in cows:
How many times per day
At what times of the day
For how long are cows observed at each time
Where are cows being observed
Who is responsible for the heat detection
If the person responsible for heat detection is unpaid, what are the
other responsibilities of this person?

 

 

 

 

96

D3. lf visual heat detection without aides is being utilized in heifers:
How many times per day
At what times of the day
For how long are heifers observed at each time
Where are heifers being observed
Who is responsible for the heat detection
If the person responsible for heat detection is unpaid, what are the
other responsibilities of this person?

 

 

 

 

 

 

I Synchronization Programs — Part E

 

E1. Were any synchronization programs being used on your farm in 2005?

Cows Heifers
I I Yes (Skip to E4) I I Yes (Skip to E4)
CI No (Skip to and answer E2) [:1 No (Skip to and answer E3)

Please note: If you answered ‘No’ to utilization of synchronization
programs in both heifers and cows, please skip ahead to Part F.

E2. lf synchronization programs were ggt utilized in 2005 for cows, please
check all reasons that apply:

CI Synchronization protocols too expensive to use

[I Prefer to breed cows to a visually detected estrus

CI Inadequate facilities to restrain cows for injections

El Lack management time required to manage a synchronization program
D Not convinced of benefits of synchronization programs

[I Poor conception rate to timed artificial insemination

C] Other (Please Specify)

 

E3. If synchronization programs were n_ot utilized in 2005 for heifers, please
check all reasons that apply:

 

1:] Synchronization protocols too expensive to use

E] Poor response of heifers to synchronization protocols

El Prefer to breed heifers to a visually detected estrus

El Heifers are at an inconvenient location

CI Lack of handling facilities for heifers

[:l Lack management time required to manage a synchronization program
El Not convinced of benefits of synchronization programs

[3 Poor conception rate to times artificial insemination

1:] Other (Please Specify)

 

97

E4. lf synchronization programs were being used in cows and/or heifers in 2005,
please select the reasons for use below. (Check all that apply,)

E] Setting up cows/heifers for first postpartum Al service
[:1 Resynchronization for 2"d or greater service

[I Synchronizing and breeding problem breeders

[I Breeding cows/heifers with ovarian cysts

El Breeding anestrus/anovular cows/heifers

C] Other (Please Specify)

 

E5. If synchronization programs were used in cows and/or heifers in 2005,
please select those that were used in your operation. Please note any changes
from the described protocols in the margins.

Cows Heifers

CI Ovsynch I] Ovsynch
% cows % heifers

D Presynch [I Presynch
% cows % heifers

CI Cosynch El Cosynch
% cows % heifers

[:I Heatsynch
% cows

[I ClDR with PGan
% cows

El Targeted Breeding Protocol
% cows

[I Use of a single injection

of PGFZGI to bring lactating

, cows into estrus for Al

[I Use of a timed Al in lactating
cows after a single injection of PGFZ,Jl

|:l Other (Please Specify)

 

98

[:I Heatsynch
% heifers

El ClDR with P650
% heifers

CI Targeted Breeding Protocol
% heifers

CI Single injection of PGFz.1| for
synchronizing estrus

[:I Two injections of PGan
administered at 11-14 day
intervals

CI Melengestrol acetate (MGA)
combined with PGan

D Other (Please Specify)

 

E6. If synchronization programs were utilized on your dairy in 2005, were cows
and/or heifers monitored for estrus and inseminated between synchronization
intervals?

[I Yes
I] No

E7. What was the average cost per dose of the following items which you
utilized in synchronization programs?
Did Not
Use

 

C
In
0
O.

Costldose

 

GnRH

 

PGFZQ

 

ECP

 

MGA

 

ClDR

 

Other

DDDDDDI
DDDDDD

 

(Please Specify)

 

E8. lf synchronization programs involving injections were utilized in 2005, what
facilities were utilized for giving shots? (If more than one type of facility is
utilized, please state all facilities used.)

 

E8a. Please specify the amount of time per cow needed to give a shot
using the above facilities.

 

E8b. Who was responsible for administering the shots/program?

 

99

 

I Reproduction Sirmmary — Part F

 

F1. Please fill in the following table referring to conception rates, heat detection
rates, and services per conception in heifers and cows. Please label each group
column according to the program/method used. For example, there may be two
groups of heifers — one group receiving visual heat detection and one group
using ClDRs. Each of these groups would be labeled under heifers and their
respective conception rates, heat detection rates, and services per conception
reported.

 

 

 

 

 

 

 

 

 

 

Example Heifers Cows
All All Group Group All Group Group

Heifers 1 2 1 2

Program] Visual

Method Heat

Used Detection

Heat 65%

detection

rate

Conception 58%

rate

(all

services)

Services 1.72

per

concepfion

 

 

 

 

 

 

F2. When was the last major change in your reproduction program?

 

F3. What was the last major change in your reproduction program?

 

 

 

Why was the above change in your reproduction program made?
I] Herd expansion
C] To remedy reproductive performance
[3 Advice of management team
[I New/different facilities
I:I Other (Please Specify)

 

100

APPENDIX 2: SYNCHRONIZATION PROGRAM REFERENCE SHEET
INCLUDED WITH SURVEY

 

 

Prostaglandin F211 (PGFZa)

-) Common commercial products include Lutalyse, Estrumate, Prostomate
Gonadotropin Releasing Hormone (GnRH)

9 Common commercial products include Cystorelin, Fertagyl, Factrel
Estradiol Cypionate (ECP) — Long acting estrogen
CIDR— intravaginal progesterone insert
Melengestrol acetate (MGA) - progestin that suppresses heat and prevents ovulation

 

 

Ovsynch (Does not include PGF 2a injections before the first GnRH injection)

PGF 211
GnRH l GnRH
I l I Timed AI

 

I 7 days I2 days W - 24 hours

Presynch (May include an additional PGFZa injection 14 days before the first PGF 2a

 

injection)
' PGan PGF2a PGan
GnRH GnRH
If ll I ll I Timed Al
14 days 14 days I 7 days 2 days I 0 - 24 hours >

 

 

 

Heatsynch (Modification of either the Ovsynch or Presynch protocols illustrated above
in which ECP is used in place of the second GnRH injection as the

 

 

ovulatory stimulus)
Timed Al
PGF2a PGFZa PGF2a at 43 h or
GnRH ECP breed to
estrus

v I
g 1 1 1 g

I 14 days I 14 days r 7 days 1 d I

101

Cosynch (Specific form of Ovsynch in which breeding occurs concurrently with the
second injection of GnRH. )

Timed Al concurrent with

GnRH shot
PGF2a
GnRH I GnRH

l l |
I 7 days I 2 days I

V

CIDR with PGF2a (The CIDR is inserted on day I, followed by a PGF 2a shot on day 6,
and removal of the CIDR on day 7. Insemination occurs on detected estrus following
CIDR removal.)

PGF2a
Insert CIDR Remove
ClDR Al on
I I detected
L I I GStI'us A

 

| 5days I1day I0—48hours

Targeted Breeding Protocol (PGF2a injections are given 14 days apart and
inseminations occur on detected estrus after the second and third injection. When estrus
is not detected after the third injection, one timed AI can be given 72—80 hours after the
third injection.)

 

 

 

PGF2a PGF2a PGF2a
Timed Al if not bred to
it it detected estrus
l r
I 14 days 14 days 72 — 80 hours '

 

Breed upon detected estrus following
2"" and 3rd shots

 

 

 

102

Melengestrol acetate (MGA) combined with PGF2a (Oral feeding of MGA at
.5mg/head/day for 14 days and then fed no MGA for the next 19 days. An injection of
PGF2a on day 33 is administered. Breed heifers showing heats beginning 24 hours post
PGF 2a, and used timed A1 at 72 hours post PGF2a shot for those not showing heats.)

 

 

PGF2a
Timed AI at 72 hours if
I Feed MGA I Feed No MGA 11 "°‘ ”'9“ ‘° ”W5
I 14 days | 19 days 72 hours I

103

REFERENCES

104

REFERENCES

Britt, J. H. 1985. Enhanced reproduction and its economics implications. J. Dairy Sci.
68:1585-1592.

Caraviello, D. Z., Wiegel, K. A., Fricke, P. M., Wiltbank, M. C., F lorent, M. J ., Cook, N.
B., Nordlund. K. V., Zwald, N. R., and C. L. Rawson. 2006. Survey of

management practices on reproductive performance of dairy cattle on large US
commercial farms. J. Dairy Sci. 89:4723-4735.

De Vries, A. 2006. Economic value of a pregnancy in dairy cattle. J. Dairy Sci. 89:3876-
3885.

Fricke, P. M. 2003. Monitoring reproduction from the starting gate. Proceedings of the 6th
Western Dairy Management Conference. pg 77-88.

F ricke, P. and D. Niles. 2003. Don’t fall for the bulls-are-cheaper, more-fertile myth.
Hoards Dairyman. Oct 25:8.

Groenendaal, H., D. T. Galligan, and H. A. Mulder. 2004. An economic spreadsheet
model to determine optimal breeding and replacement decisions for dairy cattle.
J. Dairy Sci. 87:2146-2157.

Hadley, G. L., S. B. Harsh, and C. A. Wolf. 2002. Managerial and financial implications
of major dairy farm expansions in Michigan and Wisconsin. J. Dairy Sci.
85:2053-2064.

Hadley, G. L., C. A. Wolf, and S. B. Harsh. 2006. Dairy cattle culling patterns,
explanations, and implications. J. Dairy Sci. 89:2286-2296.

Harris, T. R., and H. P. Mapp. 1986. A stochastic dominance comparison of water-
conserving irrigation strategies. Am. J. Agric. Econ 68:2 298-305.

105

Johnson, D. 2005. Reduced input dairy farming may be an option.
http://www.extension.umn.edu/dairy/dairystar/02-26-05-Johnson.htm Accessed
June 7, 2007.

Klemme, R. M. 1985. A stochastic dominance comparison of reduced tillage systems in
corn and soybean production under risk. Am. J. Agric. Econ. 67:550.

Lucy, M. C. 2001. Reproductive loss in high-producing dairy cattle: where will it end? J.
Dairy Sci. 84:1277-1293.

Mas-Collel, A., M. D. Whinston, and J. R. Green. 1995. Microeconomic Theory. Oxford
University Press, Inc.

McCarl, C. A. 1990. Generalized stochastic dominance: An empirical examination.
Southern Journal of Agricultural Economics. 49-56.

McGuckin, T. 1983. Alfalfa management strategies for a Wisconsin dairy farm: an
application of stochastic dominance. North Central Journal of Agricultural
Economics. 5:1 43-49.

Meadows, C., Rajala-Schultz, P. J ., and G. S. Frazer. 2005. A spreadsheet-based model
demonstrating the nonuniform economic effects of varying reproductive
performance in Ohio dairy herds. J. Dairy Sci. 88:1244-1254.

Meyer, J. 1977. Choice Among Distributions. J. Econ. Theory. 14:326-336.

NASS (National Agricultural Statistics Service. 2007. Quick Stats (Dairy) U.S. & All
States Data — Dairy. Milk Cows by Size Groups: Operations.
http://www.nass.usda.gov/QuickStats/ Accessed June 15, 2006

Nebel, R. L. and S. M. Jobst. 1998. Evaluation of systematic breeding programs for
lactating dairy cows: a review. J. Dairy Sci. 81:1169-1174.

Nebel, R. L. and S. M. Jobst. 1998. Symposium: Gonadotropin-releasing hormone and
prostaglandin for estrus detection. J. Dairy Sci. 81: 1169-1174.

106

0‘ '
‘L‘I' I

in}! ur

min .
aired"!

Nebel, R. L. 2002. Subject: What should your AI conception rate be?
http://www.ext.vt.edu/news/periodicals/dairy/2002-O5/aiconception.html
Accessed March 19, 2008.

Olynk, N. J. and C. A. Wolf. 2007. Expected net present value of pure and mixed sexed
semen artificial insemination strategies in dairy heifers. J. Dairy Sci. 90:2569-
2576.

Overton, M. Theriogenology. 2005. Cost comparison of natural service sires and artificial
insemination for dairy cattle reproductive management. Theriogenology. 64:589-
609.

Pursley, J. R., M. R. Kosorok, and M. C. Wiltbank. 1997. Reproductive management of
lactating dairy cows using synchronization of ovulation. J. Dairy Sci. 80:301-306.

Pursley, J. R., M. O. Mee, and M. C. Wiltbank. 1995. Sychnronization of ovulation in
dairy cows using PGF2a and GnRH. Theriogenology 44:915.

Pursley, J. R., R. W. Silcox, and M. C. Wiltbank. 1998. Effect of time of artificial
insemination on pregnancy rates, calving rates, pregnancy loss, and gender ratio

after synchronization of ovulation in lactating dairy cows. J. Dairy Sci. 81:2139-
2144.

Seidel, G. E., Jr., and J. L. Schenk. 2002. Field trials with sexed, frozen bovine semen.
Pages 64-49 in Proc.l9th Technol. Conf. Artif. Insemination Reprod. Natl. Assoc.
Anim. Breeders, Columbia, MO.

Seidel, G. E., Jr., J. L. Schenk, L. A. Henrickhoff, S. P. Doyle, Z. Brink, R. D. Green, and
D. G. Cran. 1999. Insemination of heifers with sexed sperm. T heriogenology
52:1407-1420.

Stevenson, J. 2003. Why natural service is not the solution. Hoards Dairyman.
Supplement, Oct. 25:6

107

 

Tenhagen, B. --A., M. Drillich, R. Surholt, and W. Heuwieser. 2004. Comparison of
timed AI after synchronized ovulation to A1 at estrus: reproductive and economic
considerations. J. Dairy Sci. 87:85-94.

Thomas, C. V., M. A. DeLorenzo, D. R. Bray, R. N. Weldon, R. A. Bucklin, and J. G.
Martin, III. 1997. A stochastic economic analysis of large herringbone and
parallel milking parlors. J. Dairy Sci. 80:2418-2428.

Wolf, C. A. 1999. Analyzing reproductive management strategies on dairy farms.
Michigan State University Agricultural Economics Staff Paper 99.

Wolf, C., S. Hanson, E. Wittenberg, and S. Harsh. 2002. Discount rates for dairy farms.
Michigan State University Agricultural Economics Staff Paper 02-08.

Zacharias, T. P., and A. H. Grube. 1984. An economic evaluation of weed control
methods used in combination with crop rotation: a stochastic dominance
approach. North Central J. Agric. Econ. 6:113.

Zwald, N. 2003. Does the extra effort of AI pay off‘? Supplement, Hoards Dairyman. Oct
25:11.

108

    

MlCHl Al E UI V R LIBRARI S

llIIIIIIIIIIIIIIIIIIIIIIIIIII