AGBM320:

Introduction

In this lesson, you'll begin connecting the dots to analyze the agricultural products market. First, I will illustrate some of the available free resources that a market analyst can use to find useful information and data to analyze the market for a specific commodity. Second, I will focus on simple, univariate and bivariate statistics in an attempt to characterize the distribution of data series that can be downloaded from publicly available sources. I will also emphasize the usefulness of simple graphical analysis tools, including scree plots, line carts, and histograms, in order for you to learn about data behavior. After this lesson, you will understand how a properly made graph enables you to say a lot about a data series, because as the old saying goes, a picture is worth a thousand words.

Learning Objectives

At completing this lesson, you should be able to

• find relevant data sources from publicly available resources for the analysis of agricultural markets;
• retrieve, import, and manipulate acquired data into Microsoft Excel; and
• use univariate and graphical analyses to perform preliminary data analyses.

Lesson Readings & Activities

By the end of this lesson, make sure you have completed the readings and activities found in the Lesson 2 Course Schedule.

Data and Tools: How to Perform Market Analysis

In this lesson, I will focus on providing an overview of the types of data and basic tools needed to analyze the market of agricultural commodities.

Before looking at the different data resources at our disposal, let’s talk for a moment about who is going to need/use what resource. In Figure 2.1, you will find a schematic representation of the different analyses that a market analyst can perform to assess changes in market trends. The data requirements, as well as the tools needed to perform each analysis, will differ depending on the type of analysis performed and the selected approach.

The first distinction to note is if the analyst focuses on changes in commodity demand and supply (fundamental analysis), or if the focus is instead on previous figures and information in order to assess future price fluctuations (historical analysis). Traders who use a fundamental analysis rely on supply and demand information, either reports/private information or data, which is analyzed using either descriptive or more advanced statistical tools. Analysts who uses historical analysis rely heavily on long data series of previous information and statistical methods (time series analysis), or alternatively, on qualitative methods based on spotting patterns on charts (technical analysis). In the explanation that follows, I will illustrate briefly how these approaches work to set the stage for the remainder of the course.

Historical Analysis

Historical analysis is based on acquiring and using information regarding past fluctuations in prices (or other economic measures) to predict future changes. When the analysis is based on long-term time series data, a statistical regression analysis (time series analysis) can be used to review small intervals in this data. In theory, a longer time series produces better results. However, the reality is that by doing so, you run the risk of including shocks in your data. Therefore, it is important to consider the time period you are using before beginning your analysis. In contrast, a technical analysis focuses on specific short-term patterns in commodity trading to predict the trajectory of economic variables—usually prices. You'll rely heavily on charts and graphs to complete this form of analysis.

Technical analysis is particularly important when it comes to futures trading and developing futures contracts. Futures contracts are tools that allow you to commit to purchasing a specific amount of a commodity at a certain price and delivery date in order to gain a profit or protect against price fluctuations. I'll discuss futures contracts and their options in more detail later in the course.

Data on Prices From NASS

As  previously mentioned, you can find agricultural commodity price data through the National Agricultural Statistical Service (NASS). The prices paid group in the economics sector on the NASS website should be one of the first places you look for price indexes for commodities and other inputs (including fertilizers, pesticides, and other production factors). This information will be vital to your time series analysis, as it generally requires a large amount of data on prices.

Let's walk through an example of retrieving NASS price information together:

1. Choose the crops sector.
2. For the group, select horticulture.
3. Set mushrooms as the commodity.
4. Under "Choose Data Items," scroll to and select "Mushrooms, Agaricus, Sales Measured in $/lb."
5. For domain, select total.
6. Set the geographic level to national.
7. Click Continue.

You will see that in the case of mushrooms, only annual prices are available using “Get Data,” which you can visualize.

If you click "More" under the statistics table, you'll be redirected to the Quick Stats.

Other commodities, such as milk or eggs, will have monthly data available at the national and state levels, which could become a limitation for your analysis. You may need to search out other sources if you require weekly or daily prices.

Other Data Sources

There are several other data sources that you should consider as a market analyst, especially when performing a fundamental analysis and using statistical tools.

• U.S. Census Bureau: Economic conditions such as income, demographics, and additional statistics are available for your use. This information is particularly useful for comparing geographic locations. The American Fact Finder page is a good place to find information on different locations.
• U.S. Bureau of Labor Statistics: Data on unemployment and price indexes are available to assess the impact of inflation on consumers and producers. Unemployment rates are available at the national, state, and county levels. Additionally, the consumer price index can provide you with additional data on consumer purchasing power over an extended period of time. You should also refer to the producer price index when collecting data on the demand for goods.

The Role of Policy in the Determination of Prices 

Changes in policies (agricultural or not) are an important determinant of market patterns and trends. Therefore, they should not be overlooked by market analysts, regardless of whether they use a fundamental or historical analysis. In a fundamental analysis, the role of policies is clear: Policies usually result in either supply or demand changes (or both). Their importance is marked although more nuanced in a historical analysis; for example, a sudden shock in demand or supply due to a change in policy (or a shock in prices) may lead to unusual fluctuations. Time series regression methods may not be able to account for these fluctuations properly, unless you are aware of their existence. As a consequence, unusual patterns in charts may be difficult to interpret regarding stakeholder behavior variations.

There are countless sources that you can rely on to acquire information on relevant policies. For example, the Farm Service Agency provides considerable information to farm operators regarding associated policy changes. Sector-specific organizations provide policy information on their sectors. One example is the International Dairy Foods Association, which discusses major dairy sector political issues.

Federal Marketing Orders

One type of policy that you should pay special attention to in your analysis is federal marketing orders. Historically, federal marketing orders function as a way to control local fluctuations in commodity prices. They do this by setting a minimum selling price, or a price floor. More recently, federal marketing orders have started to focus on standardizing production practices and providing technical support to farmers.

Simple Statistical Analysis Measures

Once your data have been collected and downloaded, you need to understand how to use the tools necessary to analyze them. In the remainder of this lesson, I am going to illustrate some simple statistical and descriptive measures that will enable you to properly describe a series of data. Regression analysis and more on economic modeling will be covered in the next lessons.

Mean: Describes the “center” of the distribution of a data series; it represents what you can expect, on average, to observe if  you randomly select one instance from the series of data.

For finding the mean, let X be a data series made by n observations (x₁,…,x_n), that is n data points. The mean of the data series (or the average of the data series, or sample average) is then defined by

$\overline{X}=\frac{{\displaystyle \sum _{i=1}^n{x}_i}}{n}$

Median: Represents the middle point of a data series. If all the observations in the data series are sorted in either ascending or descending order, the median would be that data point that separates the sorted series in two halves.

The mean and the median are also called centrality measures of a data series. If they are close to one another, the data distribution is likely to be symmetric, and it should show some well-defined properties.

Mode: The data value with the largest number of observations; this value reoccurs the most in the data series. For a data series that presents only a few values, the mode is an important indicator because it allows you to see what the most reoccuring event is.

Variance: The measure of dispersion (or variability) of the data around their mean; it is calculated as the sum of the squared differences between the individual observations and the mean of the data series, divided by the number of observations minus 1:

$s_x^2=\frac{{\displaystyle \sum _{i=1}^N{\left(X_i-\overline{X}\right)}^2}}{\left(n-1\right)}$

It should be noted that when more observations are closer to the mean, the distribution of the data is tighter, or less dispersed, and the variance is small. When many observations are far from the mean, the distribution of the data is more dispersed and the variance is large. Data points that exist far from the rest are outliers, and they could cause the variance to be very large.

Standard Deviation: Another measure of dispersion, which is calculated as the square root of the variance. One of the main advantages of using the standard deviation in place of the variance is that it is expressed in the same unit of measure as the data.

It is not uncommon to assess the dispersion of the data by considering how much of the data are “contained” within plus or minus” ( ± ) a certain number of standard deviations from the mean. It is customary to calculate intervals equal to one and two standard deviations above and below the mean. For many data series, the largest part of the observations fall within this interval. One classical example is the normal distribution.

Normal Distribution

A data series is distributed as normal (or follows a normal distribution) if

• its distribution is symmetric around its mean,
• the mean and median coincide (or mean = median),
• it is fully defined by its mean and standard deviation,
• 68% of the observations are within one standard deviation of the mean, and
• 95% of the values are within two standard deviations of the mean.

Large samples of data usually tend to be distributed normally; many consider data distributed normally to be “well behaved” because only the mean and standard deviation are needed to fully describe it

Covariance: A measure of how two variables change together is calculated by the sum of the product of the difference of each observation for a data series and its mean, divided by number of observations minus 1:

$s_{XY}=\frac{{\displaystyle \sum _{i=1}^n\left(X_i-\overline{X}\right)\left(Y_i-\overline{Y}\right)}}{\left(n-1\right)}$

Correlation Coefficient:  This is a bounded measure of how two variables change together. The correlation coefficient is calculated by dividing the covariance between two variables by the product of their standard deviations:

$r_{XY}=\frac{s_{XY}}{s_X{s}_Y}$

The main appeal of using the correlation coefficient instead of the covariance is that it is adimensional, meaning that it has no units of measurement. Since the correlation coefficent exists between -1 and 1, it allows the market analyst to quickly assess how much two variables move together along the graph. The larger the correlation coefficient (either positive or negative), the more the two variables vary together. Usually, 0.5 is considered as the rule of thumb to assess how my <<ED: Paragraph incomplete, please revise.

Video Tutorial

In the demonstration that follows, you will analyze two data series representing the U.S. domestic production of cheddar cheeses and Italian chesses to assess whether the two data series are related. Consider repeating this tutorial on your own as it will show you how to download and manipulate data from the USDA National Agricultural Statistics Services Quick Stats 2.0.

To complete the tutorial you can also access the file AGBM320_Lesson_2_complete.