Evaluating Droughts Impact on Live Cattle Prices

Ryleigh Travers and Brandon Dodd

The Question:

What is the impact of drought on cattle prices within the 5 major cattle market regions?

Who Are We?

We are a group of commodity analysts concerned with the extensive drought over the past 20 years. We would like to see if the drought index has an actual correlation with cattle prices which could improve predictions and commodity forecasts going forward. 

Who Are You?

You are producers representing each of the five major cattle market regions (Colorado, Kansas, Nebraska, Texas/Oklahoma, and Iowa/Minnesota) planning to present this information at your state cattlemen's association meetings. 

Background Information & Our Intuition

As has been observed over the years, when we experience times of high drought, we usually witness contraction within cattle inventories; and we witness these changes due to reduced access to feed. Now feed can come in the form of pasture grass or can come as a mixed ration, but the problem is, is that when these feedstuffs become limited we see increases in their relative prices as an attempt to ration demand. While normally, this is something that would work easily with most consumer products, the problem with that in relation to cattle is that it increases the market volatility producers have to deal with when putting up cattle for beef. When we increase prices to ration demand, what happens is that producers begin liquidating their herds. This is bad in the grander scheme of things not only for the fact that less cattle mean less beef in the long run, but because for cattle, their production cycle lasts about 10 years. This can make herd rebounds and rebuilding very challenging. 

Therefore, in theory, if drought persists and causes such problems, we anticipate that cattle prices would become elevated as well. Low supply of feed causes low supply of cattle which means rationing in the demand of beef as well.  

Explanation of Analysis

In assessing droughts impact on cattle prices, we will take a look at cattle prices based off of the 5 main cattle markets previously mentioned. 

1. Colorado

2. Nebraska

3. Kansas

4. Texas/Oklahoma

5. Iowa/Minnesota

The data for prices comes from the Livestock Marketing Information Center which derives these prices from reports sent out by the USDA. Additionally, in evaluating these prices, we will convert them from nominal to real, meaning we adjust them for inflation, so that we are provided with a consistent measure across time. One could say that in running a fixed effects regression, if we control for time then this step may be obsolete, however, if we want to look at the data from a general perspective that compares drought against cattle prices, then we cannot assess price overtime in the form of nominal. This would result in distorted effects from inflation that could draw away from our analysis.

We will then take a look at the drought severity and coverage index (DSCI) as our treatment on the changes and variations in cattle prices. To reiterate, our question above is assessing drought impact ON cattle price. Thus, we need to ensure that we are specifying our DSCI values as the treatments ON prices. 

In doing this, we will be assessing the marginal impacts that drought has per month on how prices inevitably shift in each region. Being there is such a difference in the weather within Minnesota compared to Texas, we have to isolate those regions and assess the changes in price through a frequent enough measure of time (monthly) that we are able to capture those fluctuations. 

Additionally, we will control for the region and time frame so that we can assess how the impact is being sustained relative to the selected times without "washing" our data. We want to ensure that the regression is considering what is occurring within each subsequent unit of time and space rather than just over all general data. 

We also in doing this want to control for cattle inventories. The idea here is that some regions contain way more cattle than others. We especially see this in the Texas/Oklahoma region when compared to the others. So, if we are potentially in an area in which their may be a surplus of animals, we want to make sure that there is no drag in price because of such patterns. Another way to think of why this is important is because when we have regions with dense populations that doesn't neccessarily mean there is the packing or feeding capacity to maintain all of those animals. Thus, some animals may need to be shipped to other regions which comes at the detriment of increasing shipping costs and decreasing the value of the animal to account for those transportation fees. 

Summary Statistics 

Regression Model

Where:

•  Price_it is a measure of live cattle prices across each region from January 2000 to March of 2023.

• Beta multiplied by DSCI_it is a measure of drought across each region from January 2000 to March 2023.

• Gamma Inv_it is a measure of cattle inventories across each region from January 2000 to March 2023.

• Delta_t is a control variable for time. 

• Alpha_i is a control variable for our regions and/or unit of space.

Here we use a fixed effects regression seeing as how we are looking at the impact of a continuous treatment variable on a continuous outcome variable. The opposite is said for difference in difference in that we would have a binary treatment. Drought indexs vary over time and space, thus the reason to use fixed effects. 

Regression

Dependent Var.:        Real_Price                 

Drought_Index    -0.0019 (0.0012)

Fixed-Effects:   ----------------

Cattle_Inventory              Yes

Region                        Yes

Date                          Yes

________________ ________________

S.E.: Clustered  by: Cattle_Inv..

Observations                1,327

R2                        0.99763

Within R2                 0.00481

---

Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

Results of Our Regression

As can be seen by the table above, we were left with a regression coefficient that states for every 1 unit increase in the value of our drought index we anticipate to see a decrease in the value of live cattle by $0.0019 per hundred weight. Now as we had previously thought, this doesn't align with our assumption. When we move down the table to our within r squared, we see that only 0.48% of the variation within each household unit is captured by the model. This overall meaning that we aren't exactly seeing what we had hoped for, and there is not a significant correlation between drought and the relevant current live cattle price. The opposite is observed, however, when we control for fixed effects to which our r squared is then adjusted to a value of 99.76%. Below we can get further into some exploratory analysis to see what's going on. 

Explanation of Map

The map above breaks out cattle inventories by region. It is important that we consider that each region has a varying levels of inventories so the Texas/Oklahoma region does not skew the results in comparison to the Colorado region. On the map, the darker color represents higher cattle inventories and the lighter color represents lower cattle inventories. Texas/Oklahoma have the highest inventory at 2.5 million head and Colorado has the lowest at 105,000 head. If we were to not control for this variability in cattle inventories then Texas/Oklahoma would have a greater impact on the final regression which would not be intended for the final result. 

Explanation of Map

The map above shows the severity of the drought based on the time of year. The more red, the better the drought index and the more green the worse the drought index. The different regions are grouped together based on their collective whole of their DCSI and the map is broken down by year, and by quarter. We decided to break it down this way because it is much more indicative to look at each specific quarter for their current state of drought as opposed to the collective whole. Typically, Q1 and Q2 would have a less severe drought due to the natural ebb and flow of weather patterns. Typically, when there was a severe drought in Q1 and Q2 it was just carrying over from a drought experienced in Q3 and Q4. What interesting about this, is when we look at the interactive scatter plots to take a look at this relationship between drought and prices, what we actually see is more of a positive relationship during wetter periods when we have a lower drought index and a negative correlation on price when drought is high. We can explain this through herd liquidation. When drought is bad and we restrict the availability of feed, we see increased input costs, and thus also the means of doing business. The resulting herd liquidation sends a surplus of cattle into the market which in turn pulls prices down. It isn't until the follow through effects of the drought that we begin to see cattle prices increase as more cows are kept back to rebuild the herd. 

Limitations 

Based off of the results of our regression analysis we're able to greater capture some of our conceptual limitations. Number one being that we cannot entirely estimate the impact that drought has on cattle prices unless we take into consideration lagged time effects. To take a step back this regression estimated one key aspect, is drought an indicator of cattle price? The condensed version is not really, not unless we see sustained drought that causes elevated market conditions over time. Why though?

The cattle market is a very complex entity unlike many other commodity markets. Life stages vary across numerous portions of the sector and understanding time frames can be difficult. When we talk about lag, we cannot simply lag all of our data by however many years or months we may speculate, but rather we have to assess lifecycle changes in the cattle market and analyze a multitude of things. What is our replacement heifer inventory? What has our calf crop been? What are placements on feed for 5, 6, 7, 8+ weight cattle? Are we having severe enough drought that there is an incentive for produces to liquidate their cows? Especially taking a look at the feeder market, we need to understand what is happening at the seedstock and cow-calf level in order to cause such liquidation from greater sustained input prices, because if we do not take a look at the younger market of animals that is setting the stage for future supplies, then we won't truly capture the effect. Again, these are things to be taken into consideration, but for the purposes of this analysis, we are not looking at these lagged effects but rather is their a market reaction to cattle prices when drought becomes present?

Additional limitations to our analysis could be the consumptive use of water. When we are looking at some regions especially, we may have access to irrigated pasture land or even just your general irrigated cropland. Thus, understanding consumption patterns for water would be useful and similarly with what snowpack totals are. Understanding snowpack means we have a greater idea of what we will have access to in irrigation water. These factors can be important seeing as how a wet winter in the mountains could ultimately make or break a regions crop development depending on what natural precipitation later occurs. 

There are many factors that affect the cattle market, so what has been discussed is not inclusive.

Conclusion

Drought is known to have many effects on the agricultural industry. Especially for crop production, access to water can make or break the industries operational patterns. When we consider crops like corn, the crops largest consumers go in the descending order of swine, poultry, dairy cattle, beef cattle, and then the remaining other parts of the industry that inevitably ends up as a part of a human consumable. So, what happens with drought can hold major implication on our animal operations. 

Earlier we briefly touched on the lagged effects that consume analytical work in the cattle industry. To reiterate, here we are looking at the relationship of the markets reaction to the value of fed cattle given the severity of drought we are witnessing. In condensed form, what we witness is that as we see an increase in the drought index by one, we see a decrease in live cattle price by $0.0019. While this contradicts our prior thoughts, we know why this may be the result and lag plays a big role. When cattle producers witness drought, they see it in the form of less available feed. This initiates the process of herd liquidation in which breeding inventories begin being dumped into the market. In the initial short-run this causes a decrease in price because of the surplus of goods, especially if demand does not change for beef. As we pull away from those drought conditions though, that's when the market starts experiencing the impacts of the prior sell off that caused cattle inventories to contract. Better weather means more feed which also means cheaper feed. This provides the incentive for producers to maintain their replacement heifers and beef cows to begin rebuilding the herd. The restricted supply of cattle being marketed for beef then causes to the price of cattle to increase. Thus this is where our lagged effects come into play, and it varies given the extent to which drought persists in a region.

There is only one way to ration demand and that's to increase prices. While our analysis did not go as planned, it provided us with a very thoughtful answer about how the production cycle works for beef cattle.  

Appendix

R Code Including an Explanation 

#To begin we need to understand what working directory we are in so that we are able to point to the proper dataset. 

getwd()

#Now we set the working directory we desire.

setwd("C:/Users/brand/OneDrive/Documents/Spring 2023/AREC 330/Project 3")

#Second we need to load in the proper packages to run our analysis.

library(readxl)

library(readr)

library(dplyr)

library(pacman)

library(fixest)

#Now in order to create a log text that contains the result of our analysis we use the sink function and provide it a lable. 

sink("feols_regression.txt")

#Here we need to call in the dataset for analysis.

Prices_Drought <- read_excel("Prices Drought.xlsx")

View(Prices_Drought) #View this data to verify its structure.

# We create a fixed effect regression model using fixest in which we specify that we are looking at real prices as our outcome, while the drought 

# index data is our treament. We then use cattle inventories, region, and date as control variables. 

m1 <- feols(Real_Price ~ Drought_Index | Cattle_Inventory + Region + Date,data = Prices_Drought)

summary(m1) #summarize the results of our regression

etable(m1) #format those regression results into an etable

sink() #Lastly to cap off the log file we use the sink function again to tell R we have included what we wish to include.

Results of Log Text

OLS estimation, Dep. Var.: Real_Price

Observations: 1,327 

Fixed-effects: Cattle_Inventory: 92,  Region: 5,  Date: 278

Standard-errors: Clustered (Cattle_Inventory) 

               Estimate Std. Error  t value Pr(>|t|) 

Drought_Index -0.001949   0.001222 -1.59464  0.11426 

---

Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

RMSE: 1.02964     Adj. R2: 0.996699

                Within R2: 0.004815

                   m1

Dependent Var.:        Real_Price

Drought_Index    -0.0019 (0.0012)

Fixed-Effects:   ----------------

Cattle_Inventory              Yes

Region                        Yes

Date                          Yes

________________ ________________

S.E.: Clustered  by: Cattle_Inv..

Observations                1,327

R2                        0.99763

Within R2                 0.00481

---

Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1