Plenary Paper pi d at the 32nd ic Co of the System Dynamics Society, Delft,
Netherlands July 20 — July 24, 2014
Modeling Government Intervention in Agricultural Commodity Markets:
U.S. Dairy Policy Under the Agricultural Act of 2014
Charles F. Nicholson
Department of Supply Chain and Information Systems, The Pennsylvania State University,
University Park, PA 16802 USA
Mark W. Stephenson
Center for Dairy Profitability, University of Wisconsin, Madison, WI 53706 USA
ABSTRACT
The U.S. Agricultural Act of 2014 creates a new “margin insurance” program under
which dairy farmers can receive indemnity payments from the U.S. government if a
margin (defined as the difference between milk prices paid to farmers and an index of feed
costs) falls below the insured level. The design of this Margin Protection Program (MPP)
suggests that it has the potential to substantially weaken feedback processes that would
adjust milk production, prices and margins if margins fall below program threshold levels,
especially if the proportion of milk covered by insurance is large. This paper describes
potential impacts of the MPP using a CLD, then uses an empirical SD commodity model
for the U.S. dairy industry based on the commodity model described in Sterman (2000) to
assess the impacts quantitatively. We compare the results of a Baseline scenario
representing status quo dairy policies to outcomes under implementation of the new MPP
during 2015 to 2018. Our analyses indicate that if margins fall to levels that activate
indemnity payments, weakened feedback processes are likely to result in persistent lower
margins, lower farm incomes and larger government expenditures than the continuation of
current policies. We also evaluate impacts under alternative assumptions about feed and
dairy market conditions, the date of the annual deadline for participation decisions and
the extent of farmer participation (proportion of milk covered by the MPP). Stochastic
simulations indicate that lower margins, decreased farm incomes and higher government
expenditures are highly probable during 2015 to 2018, but that the differences with status
quo policies are smaller with lower feed prices or higher demand for dairy products, when
participation decisions are required earlier, and when aggregate farmer participation
(proportion of milk covered) is less. These results imply that assessments of producer
decision strategies based on historical data that do not account for program impacts may
be misleading, and that participation decisions by individual producers may need to
consider the aggregated market effects of collective producer decision making.
INTRODUCTION
Commodity models based on System Dynamics have a long history beginning with
Meadows (1970), but substantive treatment of government policies in commodity market
models is less common. For many agricultural commodities, government intervention has
been an important determinant of prices and returns since the 1930s, motivated by
arguments that agriculture is inherently more risky than other business endeavors due to
the size and structure of the agricultural production businesses, asset fixity and relatively
Plenary Paper pi d at the 32nd ic Co of the System Dynamics Society, Delft,
Netherlands July 20 — July 24, 2014
few buyers. Concerns about equity were also offered as justification for government
intervention to raise farm incomes. When the programs began, average farm household
incomes were significantly lower in the U.S. than non-farm household incomes—although
this is no longer the case (USDA Economic Research Service, 2014). More recently, there
has been considerable debate regarding the appropriate role of government in commodity
market stabilization, and this debate has been particularly contentious for the U.S. dairy
industry.
Dairy prices have varied considerably during the past twenty-five years (Figure 1).
Prior to the late 1980s, however, prices were relatively stable due to the Dairy Price
Support Program (DPSP). This program offered to purchase selected dairy products at
stated prices to help maintain a desired minimum milk price (the “Support Price,” blue
line in Figure 1) for dairy farmers. These product purchase prices (and therefore the
support price) were increased during the 1970s in response to a campaign pledge by
President Carter, but were decreased during the 1980s as purchases of dairy products and
their disposal cost the U.S. government more than $2 billion in 1983. As the support
price was lowered below a market-clearing level, farm milk and dairy product price
variation increased. Industry professionals initially believed that dairy farmers and
manufacturer inventory managers would take a few years to learn how to manage this
variability, and then prices would stabilize. The history, however, indicates increasing
price variability, with changes of nearly 50% occurring between peaks and troughs (e.g.,
decrease during 2008 from above $20 to below $10 per 100 Ibs; Figure 1). In response to
ongoing variation, the U.S. government introduced additional programs that made direct
payments to farmers when prices fell below specified levels, and promoted the
development and use of risk-management tools by dairy farmers, manufacturers and dairy
product buyers. Although prices have become more variable, there is empirical evidence
for price cycles with a period of about three years (Nicholson and Stephenson, 2014),
consistent with the structure of SD commodity models (e.g., Sterman, 2000).
In 2009, farm milk prices and the margin between milk price and the costs of feed for
dairy animals fell to historically low levels. Dairy farmers in many parts of the U.S.
experienced substantial losses of business equity and many exited the industry. This event
suggested to many observers that existing dairy policies no longer provided an adequate
“safety net” for dairy farmers. Many policy options were discussed during the intervening
years, but early in 2014, the U.S. Congress passed farm legislation (the so-called “Farm
Bill”) that markedly changed the nature of U.S. dairy policy. This legislation will eliminate
the Dairy Product Price Support Program (DPPSP) that followed the DPSP and other
support policies, replacing them with a new program that provides dairy farmers with the
opportunity to purchase “margin insurance” through the Margin Protection Program
(MPP). Under this program, farmers determine a level of margin (milk price less a
specified feed cost value) they want to protect for a certain proportion of their historical
milk production, and pay premiums to the government. If average margins for two
consecutive months become lower than the level covered by the margin insurance, the
government will pay farmers an indemnity based on the difference between the observed
margin and their protected margin.
Plenary Paper pi d at the 32nd ic Co of the System Dynamics Society, Delft,
Netherlands July 20 — July 24, 2014
25
aA
$/100 Ibs
0
Jan-75 Jan-78 Jan-81 Jan-84 Jan-87 Jan-90 Jan-93 Jan-96 Jan-99 Jan-02 Jan-O5 Jan-08 Jan-11 Jan-14
support Manufacturing
Figure 1. U.S. Manufacturing Milk Price and Manufacturing Milk “Support” Price,
1975 to 2014
Given the major change in the U.S. government's approach to providing support to
dairy farmers, an ex ante analysis of program impacts is relevant. Thus, this paper has two
principal objectives:
1) Describe a causal loop diagram that provides insights into possible behaviors of the
U.S. dairy supply chain with the MPP;
2) Simulate outcomes of the MPP compared to status quo policies under different
assumptions about market conditions, participation by dairy farmers, and selected
elements program design (yet to be determined) using a detailed empirical model of
the U.S. dairy sector.
CAUSAL LOOP DIAGRAM ANALYSIS
Although the approach used in the MPP makes U.S. dairy programs more consistent
with other agricultural support programs such as crop insurance, it has several design
features that could result in the program being less effective and more costly than
expected. First, payment when margins are low will help sustain farm income, but this is
likely to prolong the periods of low prices because milk production adjustments in
response to market conditions will be muted. Second, there is evidence that the premium
payments are highly subsidized (i.e., not ‘actuarially fair’) for most of the margin levels
Plenary Paper pi d at the 32nd ic Co of the System Dynamics Society, Delft,
Netherlands July 20 — July 24, 2014
protected’, which will encourage farmers to insure larger amounts and provide insufficient
funding for indemnity payments. Third, farmers can decide for individual years whether
to insure and how much, rather than making a decision to participate over the five-year
life of the program. This could result in farmers purchasing insurance only when
payments are likely to be made, further increasing government costs. Finally, the amount
that farmers can insure could increase each year based on increases in total U.S. milk
production.
These program features suggest that if low margins occur that result in indemnity
payments, these could result in the unintended consequences of prolonged periods of low
margins and large government expenditures. The feedback processes that could result in
these outcomes include a number of key balancing and reinforcing loops, some with
relevant delays (Figure 2). To illustrate this possibility, consider an increase in feed costs
(which can comprise 50% of the variable costs of milk production). In the absence of the
margin insurance program, an increase in feed costs would reduce farm profitability,
which over time would reduce dairy farmers’ expectations of profits and they would
reduce their cow numbers (the key productive capital stock) and reduce milk per cow
(intensity of utilization of that capital stock). This would result is less milk production,
lower dairy product inventories, higher dairy product prices and higher farm milk prices.
These balancing loops (Profitability & Cows and Profitability & Productivity) suggest effects
that at least partly offset the initial increase in feed costs.
The margin insurance program alters this dynamic adjustment process by reducing
the strength of the balancing feedback implied in the Profitability & Cows and Profitability
& Productivity feedback loops by adding the Margin Profit Support loop (Figure 2). An
increase in feed costs would reduce profitability, but if it also reduces margins (i.e., milk
price less feed costs)” below the level selected by the farmer, the government makes an
indemnity payment that helps to support farm profitability, which weakens the balancing
loop that would reduce milk production. Low margins affect farmer expectations of lower
margins in the future and farmers would choose to cover larger amounts of production at
higher production levels (Margin Coverage Elected loop). If the program is sufficiently
subsidized, aggregate milk production could actually increase over time, which would
allow larger amounts of milk to be covered under the margin program in the future, also
' We evaluated indemnity payments less premium payments per farm for a wide variety of farm
configurations during 2009 to 2013, finding that the expected value of net payments was positive. This
1 subsequently.
‘ion. Farm profitability often
d,
ts also the finding from our simulations results for future years, to be dis
2p
Mist
is meas
iil to distinguish between profitability and margin in this di.
d.as Net Farm Operating Income (NFOD, which comprises revenues less variable costs (
labor, utilities, etc.). The ‘margin’ used in the program is milk price less a standardized measure of
Seed costs, which differs from profitability because it is a value per 100 lbs, because it does not consider
an individual farm’s actual feed costs and because it does not include other costs such as labor and
We assume that farmer decisions depend on NFOI, but the program operates based on
ST
utilities.
margin.
Plenary Paper pi d at the 32nd
ional Col
of the System Dy Society, Delft,
July 20 — July 24, 2014
Desired Cow
Numbers
Profitability
&
Culling
Rate
Milk Production
Per Cow
Productivity
Net Government
Expenditures
,
Farm Margin Insurance “a
Profitability
Profitabiilty
& Cows
Payments
'
Farmer Premium
a Payments
J ! '
Margin Covered
by Insurance
‘Milk Production
Program Unit
Milk Production
Margin
Covered by Insurance
Farm Milk Price
i
Margin '
Coverage
Elected
Feed Costs
Demand
Balance
Coverage
Dairy Product
Demai
Milk Production for
Which Margins Can be
Covered
Figure 2. Feedback Structure Related to Dairy Production, Demand and the Margin Protection Program of the Agricultural
Act of 2014
increasing the milk production covered by insurance (Milk Production and Allowed
Coverage loop). Although farmer premium payments will also increase as higher levels of
insurance are selected, the subsidization of the program implies that net government
expenditures would increase. Under certain conditions, it is possible that the feedback
structure implied by farmer decisions and margin program insurance design could “lock-
in” low margins, low milk prices and high government expenditures. (Although this is
undesirable for government and farmers, consumers in the U.S. and countries to which we
export dairy products would be beneficiaries of the program.)
As noted by Sterman (2000), conceptual models such as the one described above are
useful but are complemented by the development of empirical simulation models. The
extent to which the MPP would result in extended periods of low prices, low margins, low
farm incomes and large government expenditures will depend on a variety of factors that
are best assessed with an empirical model.
EMPIRICAL U.S. DAIRY SUPPLY CHAIN MODEL METHODS AND DATA
Our assessment of the impacts of the MPP uses a detailed empirical SD model of the
U.S. dairy supply chain adapted from the commodity supply chain model described in
Sterman (2000), which builds on an initial formulation by Meadows (1970). This model
has been developed and adapted to the U.S. dairy industry during the past 10 years, and
the feedback structure relevant for this analysis was discussed above (Figure 2).
Additional model details are provided in Nicholson and Fiddaman (2003), Nicholson and
Kaiser (2008) and Nicholson and Stephenson (2010). The base data used for the model
are for 2011. The model is more detailed than many SD models in part because detail
was required to capture factors considered important by industry decision makers and to
adequately represent current and future dairy policies.
The model calculates monthly outcomes from 2012 to the end of 2019 (when the
current farm legislation will be revisited). The model comprises modules that represent
farm milk supply, farm milk pricing, dairy product processing, inventory management and
trade, and dairy policies (both those existing prior to implementation of the Agricultural
Act of 2014 and the margin insurance to be implemented going forward). Each of these is
discussed in detail below.
Farm Milk Supply
The milk supply components of the model are based on four farm-size categories
based on numbers of cows owned for two U.S. regions, California and the rest of the U.S.
For each farm size category, the total number of farms is modeled, as is the average
financial situation (both elements of the income statement and the balance sheet) for each
farm category. The cost structure of farms in the different herd size categories is different
as is the responsiveness to price signals. Based on genetic improvement rates over the
past 20 years, milk per cow is assumed to grow at a potential rate of 2% per year, but is
* California is modeled separately because it is the largest milk producing state and maintains a state-
level system of milk price regulation different from the rest of the U.S.
adjusted in the short run based on the margin between farm milk prices and feed prices.
The number of cows for each farm size category is treated as a productive asset, and
modeled using an “anchoring and adjustment” approach based on Sterman (2000). This
anchoring and adjustment mechanism assumes that desired cow numbers for each farm
size category respond to the profitability (measured in terms of Net Farm Operating
Income, NFOI, which equals total revenues less variable costs for feed, labor, and other
expenses) relative to a benchmark but are based on current cow numbers. When the
desired number of cows changes, the voluntary culling rate is adjusted. Changes in the
culling rate in response to profitability changes are asymmetric: producers are assumed to
respond more fully when lower culling rates (to increase cow numbers) than to increase
culling rates (to decrease cow numbers).
Farm Milk Pricing
The U.S. government and many states maintain regulations that set minimum
allowable farm milk prices based on market prices of dairy product prices and the product
for which the farm milk is used. The details are provided in Nicholson and Stephenson
(2010) and are not discussed here because these programs will not be modified under the
Agricultural Act of 2014. Milk prices affect both milk per cow and NFOI and therefore
influence cow numbers. A standard measure of the farm milk price is the “All-milk” price
reported for the entire U.S. (including California) by the National Agricultural Statistics
Service, and this is included in the model as a benchmark price.
Dairy Processing
The dairy processing component of the dynamic model incorporates 21 products, 18
of which are “final” products (have explicit demand curves) and 13 of which are
“intermediate” products that are used in the manufacture of other dairy products (Table 1).
Non-storable products (fluid, yogurt, ice cream and cottage cheese) are assumed
manufactured in the month in which they are consumed. Storable products have
inventories, and inventories relative to sales (inventory coverage) is used in setting prices
for these products. Milk is allocated preferentially to fluid, soft and cheese manufacturing,
with the remaining milk allocated to nonfat dry milk (NDM) and butter manufacture. The
model explicitly tracks skim milk and cream quantities to ensure component (mass)
balance. To represent potential substitutability among intermediate products as relative
prices change, the lowest cost of three potential ingredient combinations (for example,
NDM versus milk protein concentrates (MPC) used in cheese manufacturing) is calculated
and adjustments in intermediate product use occur over the course of a month following a
change in the lowest-cost combination. The proportional utilization of existing
manufacturing capacity for storable products depends on current profit margins,
calculated on an aggregated enterprise basis. The manufacturing capacity for each region
was assigned based on production shares in California and the U.S. in 2011. Capacity for
cheese and whey products changes over time in response to long-term changes in
profitability.
Table 1. Dairy Product Categories Included in the Dynamic Model
Product Category Product Category
Fluid Milk Dry Whey
Yogurt Whey Protein Concentrate 34% Protein
Frozen Desserts Whey Protein Concentrate 80% Protein
Cottage Cheese Lactose
American Cheese Butter
Other Cheese Nonfat Dry Milk
Fluid Whey Condensed Skim Milk
Separated Whey Other aaa ial & Dry
Whey Cream Casein & Milk Protein Concentrates
Dairy Product Demand
Dairy product demand for final products is represented separately for California and
the rest of the U.S. Fluid milk consumption is based on fluid utilization from California
and sales from the Federal regulatory bodies that determine minimum regulated farm milk
prices using data for 2011. Consumption of other products was calculated as national
U.S. commercial disappearance (production + imports — exports — dairy industry use) and
allocated on the basis of regional population. The impacts of product prices on demand
are modeled using constant elasticity demand functions‘, which also are assumed to shift
over time in response to population and income growth. Intermediate product demand
isdetermined by the use of dairy components in the production of other dairy products,
based on relative costs. Cross-price effects for intermediate products are included for
NDM, MPC products, casein products and whey products but not for others. The quantity
demanded adjusts over time in response to price changes, rather than instantaneously.
Retail prices for fluid milk products, yogurt, cottage cheese and ice cream are modeled
using constant proportional mark-ups. Wholesale prices for storable products, as noted
earlier, depend on inventory coverage.
”
Ln . _ - ar | 2,
These constant elasticity demand functions have the basic form QD, = QD” « ~| , where
QD, ts the quantity demanded of product p, P is the relevant price per unit for product p (S/100 lbs),
" indicates a reference value used to initialize the model for QD and P, and nis the demand
elasticity ()<0). For some p, the demand also includes cross-product effects. Growth that shifis the
demand over time is included in the model formulation but not shown above for simplicity.
Dairy Product Trade
The model includes a detailed trade component. Imports and exports are
represented for 12 “tradable” U.S. dairy products. Imports and exports are modeled
separately and “net exports” (exports minus imports) can be calculated. For U.S. imports,
products are subject to Tariff Rate Quota (TRQ) and “over-quota” restrictions. The TRQ
specify a total annual amount of allowable imports at a relatively low tariff rate. We have
ignored the country-specific restrictions associated with some imported products. “Over-
quota” imports are not limited in quantity but face higher tariff rates. Both ad valorem
(percentage based on value) and specific (per unit) tariffs are represented for both
categories of imports. U.S. exports of dairy products are modeled using a simplified “Rest
of World” (ROW) that has production and inventories of tradable products but also
demands U.S. dairy products. The model uses 2011 U.S. trade data as base, and imports
and exports in future years are determined based on the growth in demand in the ROW,
relative prices in the U.S. and the world market (using Oceania pricing as a base) and
import restrictions. Total exports for each product are calculated based on interactions
between an aggregated U.S. market and the ROW, and sales for California and the rest of
the U.S. are assigned proportional to production in each region.
Dairy Policies
All current national dairy policies in addition to trade policy are represented in the
model, including the Dairy Product Price Support Program, Milk Income Loss Contracts
(MILC, a direct payments program based on milk and feed prices), and minimum farm
milk price regulation under what are called milk marketing orders. The Dairy Export
Incentive Program (DEIP, an export subsidy program) is assumed to operate under current
limits when U.S. prices are higher than world prices. Although many of these policies will
be eliminated when the Agricultural Act of 2014 is fully implemented, they are included
to represent periods before implementation and as part of a Baseline scenario that
simulates the policy status quo.
The Margin Protection Program
We modify the policy structure of the model to account for the major impacts of the MPP.
The program includes a premium schedule (Table 2) based on the margin level protected,
from $4 to $8 per 100 Ibs of milk? produced. Premiums are lower for the first tier (for
coverage on up to 4 million Ibs milk produced per year, or the production from about 180
cows) than for the second tier, so larger farms that want to protect more than 4 million Ibs
of milk will pay higher average rates. The premium payments schedules are represented
as LOOKUP functions. Although the formal administrative procedures are still being
drafted by the U.S. Department of Agriculture (USDA), it is likely that the program will
allow dairy farmers to select a participation level prior to the beginning of each calendar
year. The extent to which farmers will participate in a new program such as margin
insurance is challenging to model, but we initially assume a simplistic decision rule
*In the U.S. milk is priced in dollars per 100 lbs (“hundredweight”, abbreviated “cwt”), approximately
equivalent to 45.4 kg.
consistent with earlier assessments of the degree to which premiums are subsidized by the
government. We assume that producers use extrapolative expectations (Sterman, 2000) to
assess likely margins during the year for which the decision is to be made, and make a
decision about their degree of participation based on their expectations of margin just
prior to the beginning of the calendar year, and then explore the impact of this assumption
with additional analysis® (Table 3). The initial decision rules assume that farmers will sign
up to insure either 75% or 90% of their milk (the maximum under the program is 90% of
an historical base, which is updated each year), but the margin level protected will vary
from $4 (when expected margin is above $8) to $8 (when expected margin is below $4).
As an example that corresponds to the second row of Table 3, we assume that if farmers
expect the margin to be $6.00/cwt during the covered year, they will choose to cover 75%
of their production history with a margin protection of $6.50/cwt. Overall, this schedule
is consistent with farmers attempting to maximize benefits from a subsidized program, in
contrast with more typical risk-management decisions’. The legislation for the MPP also
includes a demand enhancement component, authorizing the U.S. government to
purchase product when margins are below $4.00/100 Ibs of milk. We assume that under
this condition the government would purchase cheddar cheese (a product that can be
purchased under current price support programs) sufficient to increase the margin to $4.00
over a two-month period.
Table 2. Premium Schedule for Margin Insurance Levels, $/100 Ibs Milk
Margin Level Tier 1 (up to 4 Tier 2 (for above 4
Insured, $/100 Ibs million lbs milk million lbs milk
Milk per year) per year)
4.00 0.000 0.000
4.50 0.010 0.020
5.00 0.025 0.040
5.50 0.040 0.100
6.00 0.055 0.155
6.50 0.090 0.290
7.00 0.217 0.830
7.50 0.300 1.060
8.00 0.475 1.360
°A key issue for program design ts how far in advance farmers must make this decision. Previous
studies (Newton et al., 2013) have argued that costs will be reduced by requirii
inthe
farmers to choose six
rr more in advance of the year in which they will protect their margin.
T We explore the impact of more limited participation as an additional scenario below.
Table 3. Assumptions Regarding Farmer Participation in the Margin Protection Program
Expected Margin
Based on 5 ‘ Margin Level
Extrapolative Proportion of Milk Insured, $/100 Ibs
Expectations, Milk
$/100 Ibs Milk
Less than $4.00 90% $8.00
$4.00 to $8.00 75% $6.50
Greater than $8 90% $4.00
Data Sources
The data used to develop the parameter values for the model are from diverse
sources, including NASS publications, U.S. Census Bureau (for trade statistics) previous
modeling studies (e.g., Bishop, 2004; Pagel, 2005), other industry documents, and in some
cases, judgment of dairy industry analysts. This use of a broad range of sources is
common for dynamic simulation models, and is consistent with the three types of data
needed according to Forrester (1980): numerical, written and mental (professional
knowledge) data.
Model Evaluation
The model was evaluated using the multiple-step process proposed by Sterman
(2000), and was judged to be adequate for its stated purpose of evaluating the impacts of
the margin insurance program. The model was also subjected to various sensitivity tests to
examine the sensitivity of its results to assumptions. Sterman (2000) identifies three types
of sensitivity: numerical, behavioral, and policy. Many results were numerically
sensitive, but we did not identify any behavioral or policy sensitivity that would
undermine the model’s usefulness for this analysis.
Scenarios Analyzed and Key Variables
We simulate and compare a number of scenarios to assess the MPP and the impact
of our underlying assumptions. To illustrate empirically the basic impacts of the program,
we compare two scenarios, a Baseline that assumes continuation the current suite of U.S.
dairy programs and an MPP scenario that assumes implementation of the dairy provisions
of the Agricultural Act of 2014 in January 2015 (conditional on the other assumptions
indicated above). The principal variables of interest include the margin, farm milk prices
and government expenditures, but we also examine impacts on dairy farm incomes,
selected dairy product prices and U.S. dairy net exports.
However, the impacts of the program are likely to depend to a large extent on market
conditions. To assess how market conditions affect program impacts, we compare
outcomes with status quo dairy policies and MPP implementation for two sets of market
conditions, Limited Impacts conditions and Major Impacts conditions. The Limited
Impacts conditions assume 25% lower feed prices (and therefore a larger margin—at least
initially) beginning in May 2015 and lasting through 2018 and a 10% increase in global
demand for all dairy products that persists for 12 months beginning in May 2015. The
Major Impacts conditions assume 25% higher feed prices (and therefore a smaller
margin—at least initially) beginning in May 2015 and lasting for through 2018 and a 10%
decrease in global demand for all dairy products that persists for 12 months beginning in
May 2015. These assumptions about market conditions will have a direct impact on
margins and milk prices and therefore on the MPP impacts compared to current dairy
programs. We further explore the ranges of possible impacts with a stochastic analysis
that uses Latin hypercube sampling of a range of possible feed costs increases (-25% to
+25% through 2018 beginning in May 2015) and global demand changes (-10% to +10%
for 0 to 24 months beginning in May 2015) for N=200 simulations. Using the same
random seed for each N=200 simulations, we can develop the empirical probability
distribution of differences in outcomes between Baseline and MPP scenarios.
Producer participation will undoubtedly influence outcomes of the MPP; at a logical
extreme, if there is very limited participation, the impacts of MPP should also be limited.
We have assumed a high level of participation for our initial scenarios based on the extent
to which the premium schedule is subsidized, but many U.S. dairy farmers are not familiar
with risk management tools more generally, and the level of effort to participate is higher
than with current programs such as MILC. It is not uncommon to hear U.S. dairy
producers indicate that they will not participate in the program—although the previous
analysis suggests that there may be some significant negative financial impacts of not
doing so (such as periods of lower margins and NFOI exacerbated by the lack of
indemnity payments).
We assess the impacts of the timing of participation decisions by dairy farmers and
the extent of participation, measured by the percentage of their production history (i.e.,
milk volume) protected by the MPP. We develop two additional MPP scenarios to assess
the impacts of these assumptions. An MPP 3 Months Advance scenario assumes that
producers must make decisions regarding the margin level to be protected and the
percentage of their production history 3 months prior to the beginning of coverage (e.g.,
by 30 September 2014 for coverage that begins on 1 January 2015), but maintains the
level of producer participation assumed previously. An MPP 3 Months Advance 25%
Participation scenario assumes that when the margin is below $8.00/cwt but above
$4.00/cwt, producers will choose to cover only 25% of their production history, rather
than 75% as initially assumed. (For margins > $8.00/cwt and < $4.00/cwt, the previous
assumptions about the margin level and percentage of production history are the same as
previously.) The expected impacts of a change in the timing will depend on how margin
expectations will change during the three months between 30 September and 31
December, and how this affect margin levels and percentage of production history that
producers will choose to protect. Assuming a much lower proportion of production
history protected when margins are between $4/cwt and $8/cwt is likely to lessen the
impacts of the MPP because it weakens the Margin Coverage Elected feedback loop if
margins fall in that range, and therefore the effects of the MPP on farm profitability,
productivity and assets (cows).
RESULTS
Empirical Results of Baseline and MPP Scenarios
The simulated outcomes (Table 4, first two results columns) are largely consistent
with our hypothesis that implementation of the margin insurance program based on our
assumptions about participation has the potential to sustain low margins, low milk prices
and large government expenditures. Compared to the Baseline, the margin used to make
indemnity payments is lower under the MPP once margins become low due to a reduction
in milk prices in 2016 that is consistent with a three-year price cycle. Once the program
becomes active as a result of low margins, the program margin value only occasionally
rises above a value of $8 (Figure 3) due to increased milk production arising from the
effects of the program that weaken feedback loops that would otherwise bring about
adjustments in response to lower profitability. The average value of the program margin is
$0.91/100 Ibs milk lower from 2015 through 2018 with the MPP. The U.S. All-milk price
is also markedly lower, with an average value after program implementation of $16.07
compared to $16.98 in the Baseline (Figure 4). (To the extent that variation in milk prices
per se is considered a management challenge, the MPP has a positive effect because it
reduces the coefficient of variation by about 30%.)
Table 4. Simulated Outcomes of the Margin Protection Program During 2015-2018,
Three Baseline Scenarios and Differences Due to the Margin Protection Program
r Difference . Difference
Baseline 6 Baseline f
A . with . with
f Difference | Limited Ae Major :
Outcome Baseline 3 Limited Major
with MPP | Impacts Impacts
Case Impacts Case Impacts
MPP MPP
All-milk price, $/cwt 16.98 -0.91 14.94 oi 20.78 337)
MPP margin, $/cwt 7.40 -0.91 7.56 -0.12 8.99 -3.37
Cumulative government 02 35 14 02 01 8.4
payments, $ billion
NFOL, Medium US Farm, 76,706 | -20,292 | 76,255 | -3,251 | 150,575 | -101,412
$/farm/year
Indemnity payments,
Medium US farm, $/farm/ (0) 34,022 0 13,056 0 52595)
year
Cumulative NFOI, $ billion 19.6 -5.0 18.4 -0.4 34.0 -19.9
Cheese price, $/Ib 1.57 -0.07 1.38 -0.01 1.85 -0.23
US net exports, cheese, mil
512 +96 857 ari 3} 77 +230
Ibs/year
Once the program becomes active, the persistent low margins result in government
payments through the end of 2016 (Figure 5), and in some cases these payments reach
more than $400 million per month. The cumulative government expenditure under the
margin program (and for purchases of cheese under the demand component of the
program) total more than $3.7 billion from 2015 to 2018 (Figure 6), compared to about
$200 million simulated under current programs. Compared to historical expenditures on
dairy programs and agricultural programs more generally, $3.7 billion is large. The
Congressional Budget Office (CBO, 2014) estimated that all “commodity” provisions of
the Agricultural Act of 2014 would cost $21.4 billion during 2015 to 2019, with crop
insurance programs costing an additional $44 billion. This level of expenditures is also
large compared to the historical cost of any previous dairy program, and could indicate
that Congress would modify the program—by raising premiums and(or) lowering coverage
levels—prior to 2018.
The MPP is simulated to make decrease farm incomes, but to make them more stable,
with fewer months in which NFOI is negative. Despite average annual payments (most
occurring during the low-price period of 2016) of more than $34,000 per year for a
medium-sized U.S. dairy farm (230 cows), simulated income during 2015 to 2018 is
decreased by about $20,000 per year compared to current dairy policies (Figure 7).
However, the program provides payments during low margins (Figure 7, green dashed
line) that decrease the number of months of negative NFOI. Lower average—but more
stable—returns may be welcomed by some U.S. dairy farmers, reflecting risk-return trade-
off preferences.
Simulated cumulative NFOI for all U.S. dairy farms is nearly $5 billion lower under
the MPP scenario than the Baseline (Figure 8). Cumulative NFOI is also less variable with
the MPP compared to the Baseline, as indicated by the more or less continuous increase
in cumulative income. In contrast, the Baseline scenario indicates periods of decreasing
cumulative NFOI, which implies that at times NFOI is negative for U.S. dairy farms as a
whole. To the extent that the reduction in variability of NFOI and the risk of negative
profitability is decreased, many dairy farmers would consider the program successful
(despite its costs).
Another outcome that would be considered positive by many in the U.S. dairy
industry is the effect of the MPP on dairy product exports. The share of U.S. dairy product
exports has grown rapidly in recent years, and most policy proposals have been examined
for their impacts on dairy trade. Because the MPP reduces the cost of the major input
(milk) for dairy product manufacturers, it lowers product prices. For example, average
American (cheddar-type) cheese prices would be reduced by $0.07/lb (about 5%) and
would be more stable (Figure 9). This would increase average annual exports of U.S.
cheese by more than 18% during 2015-2018 (96 million lbs per year, Figure 10).
Jan-15 Jan-16 Jan-17 Jan-18 Jan-19
—saseline MPP
Figure 3. Simulated Value of the Margin Used to Pay Indemnities, Baseline and Margin
Protection Program Scenarios, 2015 to 2018
Jan-15 Jan-16 Jan-17 Jan-18 Jan-19
——Baseline <==MPP
Figure 4. Simulated Value of U.S. All-milk Price, Baseline and Margin Protection
Program Scenarios, 2015 to 2018
$ million / month
x
8
Jan-15
Jan-16 sah? fas
—saseline MPP
Figure 5. Si
d Value of Monthly Government I ity Payments, B
Margin Protection Program Scenarios, 2015 to 2018
Cl
er
Jan-16 Jan-17 Jan-18
——Baseline <==MPP
Jan-19
d Value of Cumulative Government | ity Prog
Figure 6. Si
Margin Protection Program, 2015-2018
and
15000
10000
$/farm/month
8
8
-5000
Baseline =—MPP == =MPP Indemnity Payments
Figure Ts Simulated Value of Monthly Net Farm Operating Income and Indemnity
Payments for a Medi ize (230 cows) U.S. Dairy Farm, Baseline and Margin
Protection Program Scenarios, 2015 to 2018
25
0
2015-01 2016-01 2017-01 2018-01
—saseline ——MPP
Figure 8. Simulated Value of Cumulative Net Farm Operating Income for All U.S. Dairy
Farms, Baseline and Margin Protection Program Scenarios, 2015 to 2018
Jan-15 Jan-16 Jan-17 Jan-18 Jan-19
——Baseline <==MPP
Figure 9. Simulated Value of U.S. American Cheese Price, Baseline and Margin
Protection Program Scenarios, 2015 to 2018
million Ibs
Jan-15 Jan-16 Jan-17 Jan-18 Jan-19
—saseline —=MPP
Figure 10. Simulated Value of Cumulative American Cheese Exports, Baseline and
Margin Protection Program Scenarios, 2015 to 2018
Impacts of the MPP With Alternative Market Conditions
Market conditions substantially affect the impacts of the MPP compared to current
dairy policies. As expected, when market conditions are much more favorable (lower
feed prices and stronger global demand) under the Limited Impacts assumptions, the
effects of the MPP on the All-milk price and margin are much smaller, with a decrease of
$0.12/cwt rather than $0.91/cwt during 2015 to 2018 (Table 4, columns 4 and 5; Figure
11a). The impacts of MPP on government expenditures compared to the baseline are
much smaller ($200 million compared to $5 billion), as are the reductions in NFOI for a
medium-sized farm and for all US dairy farms (Table 4, Figure 12a). The decrease in U.S.
cheese prices and the increase in exports due to MPP are also much smaller under more
favorable market conditions.
When market conditions are less favorable (higher feed prices and weaker global
demand in the Major Impacts assumptions) than for the initial Baseline and MPP
scenarios, the impacts of the MPP are much larger (Table 4, columns 6 and 7). The
decrease in the All-milk price and margin is more than three times larger than for our
original market condition assumptions ($3.37/cwt compared to $0.91/cwt, Figure 11b).
Government expenditures are simulated to increase by more than $8.4 billion during
2015-2018 with MPP compared to the Baseline under these market conditions. Despite
Slow
0
Jan-15
Jan-16
== =Baseline == =MPP
Jan-17
Jan-18
Impact
Jan-19
Figure 11a. Simulated Value of the Margin Used to Pay Indemnities, Limited Impact
Baseline and Margin Protection Program Scenarios Compared to Original Baseline and
MPP Scenarios, 2015 to 2018
0
Jan-15
===MPP
Jan-16
Jan-17
Ow Jan-18
Jan-
Figure 11b. Simulated Value of the Margin Used to Pay Indemnities, Major Impact
Baseline and Margin Protection Program Scenarios Compared to Original Baseline and
MPP Scenarios, 2015 to 2018
20
15000
5000
$flarm/month
-10000
aR Bascline = MPP emBascline Limited Impact <=MPP Limited impact
Figure 12a. Simulated Value of Monthly Net Farm Operating Income and Indemnity
Payments for a Medium-size (230 cows) U.S. Dairy Farm, Limited Impact Baseline and
Margin Protection Program Scenarios Compared to Original Baseline and MPP
Scenarios, 2015 to 2018
40000
$ffarm/month
-10000
-20000
10000 +
Soe Baselne == —MpP emmsaseline Major impact —=meMPP Major Impact
Figure 12b. Simulated Value of Monthly Net Farm Operating Income and Indemnity
Payments for a Medium-size (230 cows) U.S. Dairy Farm, Major Impact Baseline and
Margin Protection Program Scenarios Compared to Original Baseline and MPP
Scenarios, 2015 to 2018
21
indemnity payments averaging more than $50,000 per farm per year for a medium-sized
U.S. farm, average NFOI is reduced by more than $100,000 per year during 2015 to 2018
(Table 4, Figure 12b), and cumulative NFOI for all U.S. dairy farms is reduced by nearly
$20 billion (60% of the cumulative NFOI in the Baseline for these market conditions).
There is a major impact on U.S. cheese markets (a 13% decrease in average cheese prices)
and U.S. net exports are tripled.
Thus, market conditions can have a substantial influence of the impacts of the MPP.
However, the two market conditions simulated above assume rather extreme values for
feed costs and global demand shocks. To provide further insights about the ranges and
probabilities of possible outcomes under the MPP compared to the Baseline, we assess the
distributions generated by N=200 stochastic simulations. Unsurprisingly, the range of
possible margin values during 2015 to 2018 is large for both the Baseline and the MPP as
market condition parameters are modified (Figure 13). However, it is clear that the
distribution of margin values over time has a smaller range and a lower average value for
the MPP simulations (Figure 13b) than for the Baseline simulations (Figure 13a). This is
further quantified by comparison of the average difference in the margin (and all-milk
price) values during 2015 to 2018 for each of the N=200 stochastic simulations (Figure
14). Only 1 of 200 simulations resulted in an increase in the average margin and all-milk
price during 2015 to 2018, and the average reduction in margin or milk price was
$0.96/cwt. Well more than half of the simulations in the range of -$0.25/cwt to -
$1.25/cwt. The distribution of cumulative NFOI outcomes suggests a high probability of
reductions in that value, with more than three-quarters of the simulation values in the
range from -$1 billion to -$8 billion (Figure 15). The average reduction in cumulative
NFOI for N=200 simulations was -$5.5 billion. There also appears to be a high
probability that the MPP will increase government expenditures compared to current
programs—only 1 simulation reported a reduction in expenditures with MPP compared to
the Baseline. The average increase for N=200 simulations was $2.8 billion, based on a bi-
modal distribution with more than half of the simulations in the range of $4 billion to $7
billion (Figure 16). Thus, although the exact empirical magnitude of impacts of the MPP
program are uncertain, there appears to be a high probability of the types of impacts
predicted by the conceptual model and reported in our comparisons of the initial Baseline
and MPP scenarios.
Impacts of Program Design and Producer Participation Decisions
As noted earlier, program design decisions such as how far in advance producers
must select margins levels and percentage of their production history to cover can
influence MPP impacts. Our simulations indicate that the impacts of the MPP would be
less during 2015 to 2018 with a 3-month advance decision rule rather than one that
allowed participation decisions up to the end of the calendar year (Table 5). The impact is
less in this case because producers’ extrapolative expectations three months ahead do not
fully anticipate margins lower than $4/cwt during 2016 (despite an overall downward
trend in prices and margins). As a result, the participation decision three months out
based on our decision rules is to cover 75% of production history at a $6.50/cwt margin,
rather than the decision to cover 90% of production history at an $8/cwt margin, which is
22
Susan Er ry car rar a Tem Se
a) Stochastic Simulation Results with Baseline Assumptions
b) Stochastic Simulation Results with MPP Assumptions
Figure 13. Range of Margin Values during 2015 to 1018 for N=200 Simulations for a)
Baseline and b) Margin Protection Program Scenarios
23
“325-300-275 250-225 200-175-450 125-100 0.75 0.50 0.25 0.00 0.25
Impact on All-milk Price and Margin, $/ewt
Figure 14. Distribution of Differences in the Average All-milk Price and MPP Margin
During 2015 to 2018 Between Baseline and MPP Scenarios for N=200 Simulations with
Variable Feed Prices and ROW Demand Pulse Values
Number of Simulations
18 47 46 445 46 43-2 4k 09S
Impact on Cumulative NFO}, Billion
Figure 15. Distribution of Differences in the Cumulative Net Farm Operating Income
During 2015 to 2018 Between Baseline and MPP Scenarios for N=200 Simulations wit
Variable Feed Prices and ROW Demand Pulse Values
h
24
Number of Simulations
8
. i
a ° a 2 3 4 5 6 7 8
Impact on Government Expenditures, $Billion
Figure 16. Distribution of Differences in the Cumulative Government Expenditures
During 2015 to 2018 Between Baseline and MPP Scenarios for N=200 Simulations with
Variable Feed Prices and ROW Demand Pulse Values
the decision taken if producers could decide very close to the end of 2015. The impacts
are qualitatively and quantitatively similar to those discussed earlier (Table 5), with
reductions in the average margin value, all-milk price, NFOI and cheese prices, and
increases in government expenditures and U.S. net exports of cheese. However, the lower
participation based on decisions three months out reduces government expenditures by
more than $2 billion ($1.6 billion compared to $3.7 billion). Although the specific
impacts of the timing of producer decisions will vary depending on the evolution of
margins over time between the decision and the beginning of coverage, our analysis
suggests that this implementation decision could likely influence the outcomes of the
MPP—particularly government expenditures.
25
Table 5. Simulated Outcomes of the Margin Protection Program During 2015-2018,
Baseline and Three MPP Scenarios with Different Assumptions about Decision Timing
and Participation
Difference with
Difference with ite icra
‘ Difference with | MPP, 3 Month ayes
Outcome Baseline Decision and
MPP. Advance
Decision a5
Production
Coverage
All-milk price, $/cwt 16.98 -0.91 -0.61 -0.05
MPP margin, $/cwt 7.40 -0.91 -0.61 -0.05
Cumulative government 02 aS ia 19
payments, $ billion
NFOI, Medium US Farm, 76,706 -20,292 -16,933 -782
$/farm/year
Indemnity payments,
Medium US farm, $/farm/ 0 34,022 20,722 8,248
year
Cumulative: NEOI.& 19.6 5.0 4.1 -0.9
billion
Cheese price, $/lb 1.57 -0.07 -0.05 0.00
us net exports, cheese, 512 ors a 5
mil lbs/year
We explore the market impacts of the producer participation decision when
combined with an implementation rule that assumes a participation decision three months
in advance. Although we maintain our decision rules for expected margins > $8/cwt and
< $4/cwt, a reduction in the percentage of production history covered from 75% to 25%
at a $6.50/cwt margin when margins are between $4/cwt and $8/cwt greatly modifies the
impacts of the MPP program (Table 5). The decrease in the All-milk price and margin is
much smaller (-0.05/cwt rather than -$0.61/cwt with the three-month advance decision,
Figures 17 and 18). Government expenditures are negative during 2015 to 2018, that is,
the government is collecting more in premiums than it is paying in indemnities (Table 5
and Figures 19 and 20), and the negative impacts on NFOI are also much smaller (Table 5
and Figures 21 and 22). This suggests that program participation decisions have a
significant impact on the outcomes resulting from the MPP. Importantly, this also suggests
that an individual producer's decision to participate could depend on the collective
decisions of other producers. If overall participation in the program is limited, then the
26
negative impacts of non-participation will be less—which if perceived and used for
decision-making could lead to more limited participation (and smaller MPP impacts).
Conversely, if participation is high, the costs of non-participation are also likely to be high,
and if perceived by producers and used for decision-making, this could lead to high
participation (and larger MPP impacts).
10 Fie
aN =how
Sfewt
o
Jan-15 Jan-16 Jan-17 Jan-18 Jan-19
Baseline meMPP reese MPP 3 Month Advance == =MPP 3 Mo Advance 25%
Figure 17. Simulated Value of the Margin Used to Pay Indemnities During 2015 to 2018,
Baseline and Three Margin Protection Program Scenarios With Alternative Assumptions
about Decision Timing and Participation
27
Jan-15 Jan-16 Jan-17 Jan-18 Jan-19
me
Figure 18. Simulated Value of the All Milk Price During 2015 to 2018, Baseline and
Three Margin Protection Program Scenarios With Alternative Assumptions about
Decision Timing and Participation
500
400
300
200
100
$ milion / month
Jan-15
S.
-300
we == =MPP 3 Mo Ad
Figure 19. Simulated Value of Government Expenditures During 2015 to 2018, Baseline
and Three Margin Protection Program Scenarios With Alternative Assumptions about
Decision Timing and Participation
28
$ billion
x
6
Jan-15 Jan-16
iN
———————
“1.0
—saseline MPP sere MPP 3 Month Advance == =MPP 3 Mo Advance 25%
15 to
Figure 20. Simulated Value of Cumulative Government Expenditures During 20
2018, Baseline and Three Margin Protection Program Scenarios With Alternative
Assumptions about Decision Timing and Participation
$billion
MPP 3 Mo Ad
Figure 21, Simulated Value of Cumulative Net Farm Operating Income During 2015 to
2018, Baseline and Three Margin Protection Program Scenarios With Alternative
Assumptions about Decision Timing and Participation
29
25000
20000
15000
10000
‘$/farm/month
-10000
Baseline S=MPP sere MPP 3 Month Advance = = =MPP 3 Mo Advance 25%
Figure 22. Simulated Value of Net Farm Operating Income During 2015 to 2018,
Medium-size (230 cows) U.S. Dairy Farm, Baseline and Three Margin Protection
Program Scenarios With Alternative Assumptions about Decision Timing and
Participation
IMPLICATIONS
The foregoing conceptual and empirical analyses are largely consistent in their
assessment of MPP impacts compared to current policies, albeit with considerable
uncertainty based on a range of future market conditions under which the MPP would
operate. Despite the uncertainty inherent in the stochastic analysis, there are a number of
implications of our conceptual and empirical findings:
Use of historical margin data to make participation decisions for the future could be of
very limited usefulness and may be misleading. It is common for analysts to illustrate
the potential impacts of the MPP at the farm level using historical data (for the past 5 to
10 years) for a hypothetical farm, but this may be misleading, for at least two reasons.
First, our analyses suggest that under conditions observed during the previous decade
or so, the program would have been active on many occasions (assuming at least
moderate levels of producer participation), and the MPP probably would have
markedly altered the trajectory of future margins, prices and program participation
decisions. That is, the past with the program probably would have been very different
from the actual past observed without the program and therefore should not be used to
assess the impacts of alternative farm-level decision strategies. Second, future costs
and benefits of the program for producers will depend on current market conditions
and the degree of participation by other producers, not on the potential benefits
observed under previous years. These are not easily assessed with historical data.
* Program design for implementation will likely influence MPP outcomes. We assessed
one important design decision, the timing of producer participation decisions, and
found that this can affect the impacts of the MPP and government expenditures in
particular. We did not assess the impacts of other program design issues that must be
decided by the implementing Farm Service Agency (FSA), isuch as which price series
(advanced reporting values or final values) will be used for the margin calculation,
when premiums must be paid (we assumed continuous payment of premiums in the
foregoing analyses) and whether that will influence participation decisions and how
the premium structure will be applied based on milk production thresholds. Although
the timing of decisions is likely to have the largest impact on outcomes, these other
design decisions could affect MPP outcomes, particularly if they affect participation
decisions.
* Participation decisions have the potential to markedly affect MPP outcomes. As noted
in our analysis, lower participation implies much more limited impacts of the MPP, but
these impacts are also likely to affect participation. This suggests that it may be useful
for the implementing agency (FSA) to report aggregate participation levels during the
sign-up period (e.g., the amounts of milk protected at what margin levels), which will
be useful to producers making decisions and to futures markets for dairy products in
assessing the likely impacts of MPP.
* The dairy producer participation decision is different for MPP than for other risk-
management decisions, but may not be independent of them. We assumed high levels
of participation in our initial analyses based on the implied subsidies in the premium
schedule. Although it was marketed as a risk management tool and will perform that
function to a certain extent (paying when NFOI is low), the program differs from other
insurance programs that pay indemnities in the case of catastrophic losses. Our
analyses suggest that the MPP may be frequently active during 2015 to 2016, with
substantive impacts on margins. This will affect both the future probability of
indemnity payments and the participation decision, neither of which is typical for a
product such as fire insurance (or crop insurance). Moreover, for most risk
management products, producers would make decisions based on a careful assessment
of their costs and benefits. For a highly subsidized program such as MPP, this decision
* For example, it is uncertain the interpretation of the premium schedule for the 4 million pounds of
milk production. For example, if a producer has a 6 million pound production history and wants to
insure 50% —3 million pounds, does that mean 50% of the first 4 million at the lower premium —2
million pounds — and 50% of the above 4 million pounds —1 million pounds —at the higher premium; or
does that mean 3 million pounds at the lower level.
could focus more on how to maximize benefits from the program, given its relatively
low costs. Finally, for farmers currently using other risk-management tools, the option
for coverage under MPP could modify the best use of these tools—with aggregate
effects on the markets for risk if a sufficient number of producers substitute MPP
coverage for other risk management coverage.
CONCLUSIONS
Our analyses suggest that many of the negative effects of a margin insurance program
that weakens corrective feedback processes in milk production could occur, including
persistent periods of lower prices, lower margins and large government expenditures.
However, these results are conditioned on two key factors. The first is that cyclical
behavior in U.S. milk prices results in sufficiently low margins in 2016, thereby activating
indemnity payments under the program and preventing the adjustment process. Although
our simulated milk prices are consistent with previous price patterns, it is possible that
structural changes in producer decision-making or different future feed prices could alter
this pattern so that the margin insurance program is not frequently activated during 2015
to 2018. If this occurred, then the importance of weakening the relevant feedback loops
could be minimal, because they would not be activated. However, our stochastic
analyses suggest that these types of impacts have a high probability of occurring with the
MPP under a wide variety of market conditions. Second, we assumed a significant degree
of farmer participation in the margin based on a simple decision rule derived from
estimates of the degree to which the program is subsidized. If participation is less than
what we assumed, this could also lessen the degree to which the feedback processes are
weakened by the margin insurance program, which could markedly alter the program’s
dynamics during 2015 to 2018. Finally, program design for implementation, such as the
timing of the participation decision) is also likely to influence the magnitude (although not
the direction) of the outcomes of MPP.
ACKNOWLEDGEMENTS
The authors thank four anonymous reviewers for the International System Dynamics
Conference whose comments improved the clarity and scope of this paper.
REFERENCES
Bishop, P. M. 2004. Dairy market impacts of U.S. milk protein imports and trade policy
alternatives. Ph.D. dissertation, Cornell University.
Congressional Budget Office. 2014. Letter to the Honorable Frank Lucas, Chair of the
Committee on Agriculture Regarding Budgetary Effects of the Agricultural Act of 2014,
January 28, 2014.
Forrester, J. W. 1980. Information Sources for Modeling the National Economy. Journal
of the American Statistical Association, 75:555-574.
Meadows, D. 1970. Dynamics of Commodity Production Cycles. Wright-Allen Press.
Newton, John, Cameron S. Thraen, Marin Bozic, Mark W. Stephenson, Christopher Wolf
and Brian W. Gould. 2013. Goodlatte-Scott vs. the Dairy Security Act: Shared
Potential, Shared Concerns and Open Questions. Briefing Paper Number 13-01,
Program on Dairy Markets and Policy, April. [http://dairy.wisc.edu/pubPod/Pubs/MP-
DMAP%2 0Briefing%20Paper%201 3-01 .pdf]
Nicholson, C. F. and H. M. Kaiser. 2008. Dynamic Impacts of Generic Dairy Advertising.
Journal of Business Research, 61:1125-1135.
Nicholson, C. F. and M. W. Stephenson. 2014. Milk Price Cycles in the U.S. Dairy
Supply Chain and Their Management Implications. Program on Dairy Markets and
Policy, Working Paper 14-02, May.
Nicholson, C. F. and M. W. Stephenson. 2010. Analysis of Proposed Programs to
Mitigate Price Volatility in the U.S. Dairy Industry. September.
Nicholson, C. F. and T. Fiddaman. 2003. Dairy Policy and the Origins of Dairy Price
Volatility. Paper presented at the 21st International Conference of the System
Dynamics Society, July 20-24, 2003, New York, NY.
Pagel, Erica J. 2005. Dynamic Patterns of Change in Structure Under Different Support
Policy Regimes: An Examination of the U.S. Dairy Industry. MS Thesis, Cornell
University.
Sterman, J. D. 2000. Business Dynamics: Thinking and Modeling for a Complex World.
Boston: Irwin/McGraw Hill
USDA Economic Research Service. 2014. Farm Household Well-being.
<http:/Awww.ers.usda.gov/topics/farm-economy/farm-household-well-being/farm-
household-income-(historical).aspx#.U48fasbecdl> Accessed 4 June 2014.
