Forecasting Long-Term Commodity Prices
When long-term forecasts are needed, technical analysis methods based on price patterns alone are not adequate. This is because, in the long run, too many factors impact the price of the commodity, thereby changing its pattern. Further, you cannot depend on historical pricing patterns to continue far into the future because the factors affecting the value of the commodity can quickly change. Instead, forecasters examine the underlying factors affecting supply and demand and use their judgment to forecast price. This approach is called fundamental analysis.
In this chapter, we explain how to develop long-term forecasts for different types of commodities using fundamental analysis. In fundamental analysis, the underlying assumption is the balance between a commodity’s supply and its demand establishes its price. Fundamental analysis uses qualitative reasoning to understand supply and demand, along with statistical approaches, but ultimately it depends on judgment to forecast price.
There are eight key steps in the fundamental analysis process. The first step is to gather information on supply, demand, and price. This step involves plotting supply, demand, and price data and looking at their relationships. The next step consists of building an understanding of the basics of supply—learning the underlying factors affecting supply and estimating how supply may change in the future. Similarly, this step requires an analysis and understanding of the basics of demand—of the underlying factors affecting demand and of how demand may change in the future. After applying reasoning to qualitatively assess supply and demand, the relationships between supply, demand, and price are examined using statistical tools. The correlation evaluates the strength and direction of relationships with price. If the correlation is significant, a model to estimate price can be developed using simple linear regression. If there is a financial market for the commodity, the next step incorporates future price trends qualitatively into the analysis. All the knowledge obtained about the commodity, prices from regression models, and futures prices will be combined to develop the final forecast, using judgment.
This chapter explains each of the following steps of the fundamental analysis process:
Gathering information.
Understanding supply basics.
Identifying underlying factors affecting supply.
Understanding demand basics.
Identifying underlying factors affecting demand.
Examining correlations and developing regression models.
Considering future prices.
Developing and monitoring the final forecast.
Gathering Information
Finding specific supply, demand, and price data is one of the biggest challenges with a fundamental analysis. For some commodities, data is publicly available, but for others, that is not the case. For many traditional commodities, the U.S. government can be a source of useful information from the United States and the world. The U.S. Department of Agriculture (USDA) Economic Research Service (crops and livestock), the U.S. Energy Information Administration (oil, coal, natural gas, and electricity), and the U.S. Geological Survey (metals and other minerals) websites are excellent sources of information about supply, demand, and prices for many commodities. Some of these government agencies also publish their own forecasts. For example, the USDA publishes reports of supply and demand forecasts for agricultural commodities for a 10-year period, which is available on the USDA website.1 Industrial associations such as the International Cocoa Association, the International Coffee Organization, and the Aluminum Association are also sources of commodity information and statistics.
Table 4.1 provides a summary of some information sources and categories you can use to identify factors influencing supply and demand, such as global events and economic trends, industrial and technological changes, supplier and customer behaviors, and governmental policies. Often it is not just one event or change in the market, but rather it is the interaction of numerous conditions influencing commodity prices.
It can be difficult to find the exact data needed for a fundamental analysis, depending on the commodity. Further, having timely access to new information is especially critical. The market quickly adjusts so new information can dramatically change a forecast. Thus, if your organization does not have the internal resources to devote to doing commodity research, consider using a subscription service providing analysis from experts who focus on understanding specific commodities. Considering the internal resources needed for finding and analyzing data, the benefit of a subscription service may outweigh its costs. However, it is important to stay knowledgeable about the commodity in order to use the information provided by the subscription service to make good supply chain decisions.
Table 4.1 Market intelligence sources2
Type of resource |
Examples |
|||
Company financial research |
Securities and Exchange Commission (SEC) filings (Edgar online) |
Fortune 500 |
Annual reports |
|
Global business references |
Supplier directories |
Embassies and consulates |
Harvard Business School |
U.S. Department of Commerce |
Industry links |
North American Industry Classification System (NAIS); Standard Industrial Classification (SIC) |
Sematech |
Raw material indexes |
United Nations Standard Products and Services Code (UNSPSC) |
Commodity and labor statistics |
U.S. Bureau of Labor Statistics (BLS) |
U.S. Department of Agriculture (USDA) |
U.S. Energy Information Administration (EIA) |
U.S. Geological Survey (USGS) |
Economic indicators |
Central Intelligence Agency Factbook |
Worldtrade Organization (WtO) |
World Bank |
OANDA |
A starting point in understanding a commodity is to visually assess the relationships among supply, demand, and price. This step consists of gathering annual supply, demand, and price data for 10 to 20 years. A commodity’s total supply includes its new production, recycled materials if applicable, and its stocks. Stocks are the amount of a commodity stored in inventory. Stocks can be held by any supply chain member, such as producers, distributors, and end users.
Governments also can hold stocks; for example, the U.S. Strategic Petroleum Reserve holds oil in inventory. Imports should be included if the analysis is focused on supply within a specific country rather than on worldwide supply. For some commodities, demand is called consumption. Demand can be examined on a worldwide basis or country-by-country, in which case exports of that country should be included as one component of demand.
After gathering data, annual data are plotted for supply, demand, and price for the commodity on the same line graph. Show supply and demand on the primary vertical y-axis on the left and show price on the secondary vertical y-axis on the right. Analyze the graph, looking for patterns, relationships, and unusual shifts. Typically, price increases when the surplus of supply relative to demand decreases. As the surplus of supply relative to demand increases, price typically decreases. Are there spikes, drops, or shifts in patterns? Researching the history of the commodity is important for understanding why these changes in supply, demand, and price have occurred.
Let’s take a look at an example. Figure 4.1 shows U.S. supply, demand, and price for natural gas from 2000 to 2015.3 From 2000 to 2004 the levels of supply and demand were very close to each other. The price decrease in 2002 at an average of $2.95 per cubic ft. is attributed to an economic slowdown beginning in 2001 and continued through much of 2002. The economic recovery that followed drove natural gas prices higher.
In 2005, when hydraulic fracking technology enabled the extraction of natural gas trapped in shale rock to become economically viable, natural gas supply increased, increasing the surplus of supply over demand. However, also in 2005, two major hurricanes caused production shutdowns in the Gulf of Mexico leading to a high spike in price to $7.33 per cubic ft. in late 2005. The dramatic price drop in 2008 and 2009 is attributed to a decline in industrial demand because of a major recession. As supply from shale gas production continued to outpace demand from 2009 onward prices have continued to decline. One exception is the spike in price in 2014, which is attributed to cold winter weather because natural gas is used for heating. Once you have good understanding of the overview of supply, demand, and price for the commodity, the next step in a fundamental analysis is to learn more about the underlying factors driving supply and demand. This knowledge is used to estimate the change in supply and the change in demand during the future time period needed for forecasting.
Figure 4.1 U.S. natural gas supply, demand, and price, 2000 to 2015
Source: U.S. Energy Information Administration.4
Understanding Supply Basics
This step begins with focusing on a commodity’s supply market. Who are the participants in the commodity’s supply network and what are their roles? Depending on the commodity there may be many tiers in a commodity’s supply network. The natural gas supply network consists of producers who extract natural gas using traditional wells or from shale rock; processors who remove water and impurities to convert natural gas so it is ready for use; and pipeline operators who transport natural gas from production to users, storage operators, marketers, and local distributors. Some supply networks may have vertically integrated members who own different parts of the network and perform more than one of the key supply functions.
A bottleneck restricting capacity of any one of the key functions in a commodity’s supply network will reduce its supply. The commodity’s supply will be limited by the part of its supply network having the lowest capacity; therefore, there is a need to identify which aspect of the supply network may represent a bottleneck. For example, with natural gas, bottlenecks could be caused by the number of gas wells, the capacity of pipelines used to transport gas, or the capacity of underground caverns used to store natural gas. When capacity is expanded at the bottleneck function, then the bottleneck shifts to another part of the supply network.
Further, understanding the basics of supply requires learning about how, where, and when the commodity is produced now and how production may change in the future. Several different production methods are used for some commodities, and new technologies can lead to new production methods. Typically, different production methods have different cost structures. As commodity prices increase or new production technology is developed, some production methods previously not economical become viable and can open new sources of supply. For example, natural gas was traditionally produced by drilling a vertical well for the gas to flow upward. Historically, natural gas production was determined by the number of wells. Since the mid-2000s, advances in hydraulic fracking have changed natural gas production dramatically increasing supply.5
The number of entities producing the commodity should be considered in a fundamental analysis. When there are many producers, each respective entity makes independent decisions about how much to produce based on estimated profit, considering the commodity’s expected selling price. In this situation, no single market participant has a significant influence on supply and ultimately price. Traditionally, agricultural products were in this situation since many farmers were independent operators of family farms. For some mineral commodities, because of the high cost of mining, a few firms may own the production capacity and thus can have a significant influence on supply and price.
For agricultural commodities, the potential number of acres and the characteristics of the actual crop yield affect production. Individual farmers make decisions about how many acres to plant of different crops based on profit potential. So, if in a given year, the price of a crop increases because it is in short supply, the next year farmers are likely to plant more acres, thereby increasing supply and driving down prices. In general, in developed countries, crop yield has increased significantly and continues to increase because of advances in seed, fertilizer, equipment, and technology, for example Global Positioning System (GPS). A higher yield per acre increases production.
For some metals—such as steel, copper, and aluminum—recycling from scrap, referred to as the secondary market, is an important consideration in supply. Thus, there is a need for understanding how primary markets for new production, as well as secondary markets for scrap, affect supply and, in turn, prices. As commodity prices increase, recycling can become a viable alternative to new production. For example, in the United States, recycled aluminum often exceeds new primary-market production.6 Recycled aluminum includes waste from manufacturing processes as well as aluminum used in products and returned from consumers, such as aluminum cans.
In the long run, recycling is likely to increase as the world’s demand for minerals and metals increases. Higher commodity prices make it more economical to invest in technologies for recycling. For example, historically, very little lithium was recycled. With the growth of lithium batteries in electronics and the projected growth for hybrid and electric cars, recycled lithium will be more competitively priced, and the market for recycled lithium is likely to increase.7
Location is also an important factor to consider. Production and transportation costs are often influenced by a commodity’s production location. For example, most natural gas produced in the United States is also consumed there and thus can be transported by pipeline. However, using liquid natural gas (LNG) technology, negative 260 degrees, opening up new sources of natural gas, such as Trinidad and Tobago, which were not feasible using conventional methods.8 However, with the economic and technical feasibility of shale gas, there is less focus on using LNG technology to import natural gas into the United States. Instead, LNG technology may be used for exports from the United States.
Globalization has a significant impact on supply. In the past, when agricultural commodities such as corn and soybeans were primarily produced and consumed in the same region, supply would greatly increase at harvest time and then steadily decrease throughout the year. However, many food crops are produced in both the northern and southern hemispheres. Global trade and economic ocean transport reduce the impact of the harvest season in any one region on supply and price. For example, corn is typically planted in the United States during April and May and harvested in October and November. In Brazil, the planting season ranges from October to December and harvesting from February until June, and some areas of Brazil can do two plantings a year.9
Some commodities are produced in many regions of the world while others may be limited to a specific region, depending on natural resources or climate required. When production is concentrated in a few regions, the risk of a supply disruption increases. Cocoa, for example, can only be grown in rainforest conditions occurring plus or minus 10 degrees of the equator, limiting production to a few countries. In fact, over 30 percent of the world’s cocoa is grown in a single country, Côte d’Ivoire, which has a history of political instability increasing the risk of a disruption of cocoa supply.10 In another example, Chile has over 50 percent of the world’s known lithium reserves, and thus the shift to hybrid and electric vehicles will leave the world dependent on this South American country.11
Organization of the Petroleum Exporting Countries (OPEC) is perhaps the best-known example of how a coalition of countries attempts to influence supply and price. OPEC, with its 12 member countries, has about a 40 percent market share of the world’s conventional oil supply and 60 percent of global oil trade.12 The members of OPEC meet to discuss and agree upon production quotas for oil, which impacts price. However, the differing political agendas of the member countries means reaching agreement is difficult. Further, some members do not comply with production quotas, reducing the effectiveness of the agreements.
A commodity’s stock levels, and the amount held in inventory, are part of supply. A commodity’s inventory can be held across the supply network by a number of different market participants—including producers, consumers, commodity exchanges, and governments. Commodities with seasonal harvesting patterns are typically stored, with high stock levels, immediately after harvest, and the stock levels decline with use until the next harvest. Similarly, commodities with seasonal consumption also show seasonal stock patterns. For example, demand for fuel oil and natural gas in the United States is higher in the winter because of their use for heating. During the summer, natural gas is stored in underground caverns. In the winter, natural gas is withdrawn and used as a fuel for heating.
Long-term price is also influenced by proven or anticipated reserves for commodities such as oil, natural gas, minerals, and metals. Proven reserves are those extracted using known technology. For example, OPEC claims to have almost 80 percent of the world’s proven conventional oil reserves.13 New technology can lead to discoveries of new reserves of a commodity. For example, from 2010 to 2014, the U.S. Energy Information Administration estimates of recoverable natural gas from shale increased by over 100 percent.14 However, a commodity’s proven reserves do not necessarily lead to a future increase in actual supply.
Identifying Underlying Factors Affecting Supply
After examining the basics of supply for the commodity, all the underlying factors affecting supply in the future need to be identified, as well as the extent to which you expect supply to change because of these factors. Many factors can affect a commodity’s supply including weather and climate, new technology, prices of other commodities, government policies and regulations, political instability, and input prices. Different underlying factors influence different commodities. In this section, we will discuss each of the major categories of underlying factors.
The time frame over which the factor may have an influence on supply needs to be identified for each factor. Some factors may affect supply in the short-term while others may affect supply in the long-term. Next, the impact of each factor upon supply is estimated by examining historical events and their impact on supply in the past—as well as how industry experts expect supply to change in the future.
Weather is a major factor affecting the supply of many commodities, especially agricultural crops. For crops such as corn, wheat, soybeans, and cotton, too much rain or flooding during the planting season means fewer acres can be planted, thereby reducing production. Even if a crop is successfully planted during the growing season, too much or too little rain is a problem, reducing crop yield. Of course, the weather at harvest time can also impact crop yield. In some extreme situations crops planted cannot be harvested because of the weather. In contrast, when the weather is favorable, production of agricultural commodities can exceed expectations, thereby increasing supply.
Agricultural crops are not the only types of commodities whose supply can be affected by weather. Oil and natural gas production in the Gulf of Mexico is disrupted by the weather. When major hurricanes occur, for employee safety, production platforms must shut down, reducing supply. Damage to rigs, pipelines, and refineries can halt production for a significant amount of time, as was the case after Hurricanes Katrina and Rita in 2005. Severe weather also can disrupt mining operations.
It is impossible to correctly predict in advance the impact any single weather event will have on the supply of a commodity. However, it is important to understand the climate in which commodities are produced. For example, it is probable during any hurricane season that oil and natural gas production in the Gulf of Mexico may be temporarily curtailed. In some years the chance is greater than others. The National Oceanic and Atmospheric Administration (NOAA; www.noaa.gov) is an excellent source of weather information such as drought and hurricane predictions. In fact, weather may pose a major challenge for commodity forecasting in the future. Scientists at NOAA suggest that in the United States, climate extremes have been increasing since the 1970s, and this phenomenon would be expected globally as well.15 More extreme events, such as intense storms or very high or low temperatures, means weather is likely to have a greater impact on commodity production. Factors can also be related with each other, impacting on commodity price volatility. Starting in 2007, for example, durum wheat volatility increased significantly due to unfavorable weather conditions on all the continents, the related reduction of worldwide stock levels, and trader speculation in the financial markets. Some of the companies we involved in our studies highlighted that weather risk significantly affects price volatility, but this influence is difficult to manage, due to many other factors.
Incremental improvements in technology have a major impact on a commodity’s supply over time. For example, advances in seed, fertilizer, irrigation methods, equipment, and farming practices have significantly improved crop yield for agricultural commodities. To illustrate this example, in the United States, corn yields have increased from 118 bushels per acre in 1990 to 168 bushels per acre in 2016.16 More radical changes in technology, such as deep-water drilling for oil and the ability to extract natural gas from shale, can have a major impact on supply.
Commodity prices also determine how much of a commodity will be produced. As a commodity’s price increases, there is more incentive for current producers to increase production levels and for new producers to enter the market. Of course as supply increases, if demand does not likewise increase, prices drop and over time production will also be reduced. In agricultural markets, farmers have the flexibility to shift between different crops, depending on which one they believe will bring the highest prices. Farmers face a high level of price risk and uncertainty because of unfavorable weather and thus seek to maximize their profits. For example, crops such as corn, soybeans, wheat, and cotton can be planted depending on expected prices. In another example, oil prices determine if it is economical to produce alternative liquid fuels such as biodiesel and ethanol, driving up prices for fuel crops such as soybeans and corn. Higher prices entice farmers to plant more of these crops, which when harvested, increase supply.
A producer’s decision of how much of a commodity to produce is also affected by the availability and cost of its inputs such as labor, equipment, and materials. The cost of seed and fertilizer may cause a farmer to switch to an alternative crop. Since Brazil is the world’s largest exporter of sugar, the world price of sugar is heavily influenced by their cost of production. As the cost of production increases, so does the cost of worldwide sugar.17 Globalization and the search for lower cost inputs such as labor and energy are likely to shift commodity production to low-cost countries when possible. Governments use subsidies and tariffs to retain local production.
The cost and time required to add production capacity also influence supply. Grain crops are typically planted once per year, so once the planting decision is made it cannot be changed until the next year. For some commodities it can take years to add capacity. After planting, it takes coffee trees 3 to 4 years to bear fruit. Oil refineries and mines can take years of planning, approval, and construction before being operational.
Government regulations including environment, safety, energy usage, subsidies, and taxes can have a major effect on supply. The specific type of impact depends on the type of commodity. Government regulation or deregulation can impact the industry’s structure and the number of market participants. Regulations can also impact the time required to develop new production or delivery capacity. Some industries such as mining, natural gas, and oil face an extensive government approval process, delaying capacity additions.
With globalization, many different governments with different policies and regulations can impact supply in different ways. For example, air pollution laws taking effect in 2016 in China are likely to influence the supply of primary aluminum production.18 Another example is the Russian government’s decision not to export grain in 2010 because of a severe drought, which resulted in higher worldwide wheat prices.19
Government subsidies can have a major impact on commodity supply. Sugar is perhaps the most heavily subsidized commodity in the world with most countries, including the United States, having some type of sugar subsidy.20 Restricting supply through production quotas and import restrictions results in significantly higher prices for sugar in the United States than in the rest of the world.
Political instability and war can reduce a commodity’s supply. For example, over the last 25 years, war and political instability in the Middle East have caused periodic oil disruptions and price spikes. Besides the actual supply disruption, political instability and war increase uncertainty of future supply, typically driving up prices.
A commodity’s supply is likely to be affected in different ways by a number of these different factors. Some of them are internal to the organization or supply chain, while others are external, and are closely intertwined with the well-known PESTLE framework (Political, Economic, Social, Technological, Legal, and Environmental).
Some factors may increase supply while others may decrease supply. In doing a fundamental analysis, you need to consider the overall effect from all possible factors to determine the amount you believe supply will increase or decrease.
Let’s take natural gas as an example. According to the Natural Gas Supply Association, underlying factors affecting the supply of natural gas in the United States are the availability of inputs (skilled workforce, equipment, and pipeline capacity), government regulations (permitting time and access to land), and weather.21 However, since 2005, the number one factor affecting natural gas supply is the technology used to extract shale gas. In the long run, the U.S. Energy Information Administration projects natural gas production in the United States will increase steadily to a production rate in 2040 of over 35 trillion cubic ft. per year, a 46 percent increase over the production in 2012.22
In another example, assume you are developing a forecast for a commodity for three years from now. For the commodity assume you estimate new production technology will increase supply by 8 percent in 3 years. You estimate new government regulations will reduce supply by 3 percent. You believe other factors will not significantly impact supply in the next three years. Thus the most likely change in supply in 3 years is a 5 percent increase. If the current supply of the commodity is 100 million tons, the forecast supply in 3 years is 105 million tons.
Understanding Demand Basics
Once there is a good understanding of supply and the underlying factors affecting supply, the process is repeated for understanding the basics of demand. As with supply, U.S. government agencies track demand data for many commodities available free through websites. Other sources of demand data include industry associations and many data services charging a fee for data access. Often the term consumption is used to refer to demand.
To understand the basics of demand, key questions to ask are these: how is the commodity used; who uses the commodity; where is it being used; and what are technically viable substitutes, if any? Has demand changed over time and if so why? For example, in 2015 in the United States, the largest use of natural gas was to generate electricity, followed by industrial, residential, commercial, and other uses as shown in Figure 4.2. Examination of the major consumers of natural gas since 2001 shows residential and industrial use has declined slightly, while use of natural gas to generate electricity has increased typically replacing coal. Later in this chapter, we will discuss factors contributing to these changes.
Figure 4.2 U.S. natural gas demand by sector, 2015
Source: U.S. Energy Information Administration.23
To understand demand, it is important to understand where commodities are consumed. For example, because of rapid economic growth for years China’s demand grew for many commodities driving up prices. However, by 2014, slowing growth softened demand and prices.24
Some commodities can be used for the same purpose and are often substituted based on price. Copper and aluminum can be used for the same purpose in many applications such as electronics and construction. Aluminum, titanium, and steel can be substituted for each other in some applications. High-fructose corn syrup and sugar can be substituted in many food products. Natural gas and coal can be substituted for each other in electricity generation. For your commodity, it is important to understand the technical challenges associated with using substitute commodities, as well as the effects that switching by competitors and firms in other industries may have on demand and, subsequently, on prices.
Identifying Underlying Factors Affecting Demand
Once there is a basic understanding of the factors affecting demand, all the underlying factors affecting demand in the future need to be identified as well as the extent to which demand may change because of these factors. In the long run, population growth, demographics, and economic development and growth affect demand. In addition, many factors affecting supply also affect demand. For example, weather and climate, prices of other commodities, government policies and regulations, new technologies, and customer tastes influence demand.
The world’s population continues to grow resulting in increasing demand for commodities, especially in the areas of food, energy, and metals. The United Nations projects the world’s population to grow from 7.3 billion in October 2015 to over 11 billion by 2100.25 Populations are projected to grow the fastest in developing countries, especially in India and some countries in Africa. Over the long-term, the increased demand from population growth is likely to increase prices, especially for commodities such as oil and metals having limited reserves. Of course, demand for renewable commodities such as food and fiber also will increase with population growth.
While parts of the world that are growing will have younger populations, others will experience aging populations. Demographics affect demand for some commodities. For example, with aging populations, for health reasons, dietary patterns shift to reduce consumption of some commodities such as sugar.
Economic development and growth affect demand for commodities. In developing countries, as construction for homes, businesses, and infrastructure increases, so does the demand for commodities. Economic growth in manufacturing creates a need for commodity inputs. Economic growth and the development of a middle class increase demand for materials to produce consumer products. For example, demand for wool has increased because of demand from Chinese businessmen for tailored men’s suits.26
Weather and climate affect demand for energy commodities. Harsh winters increase the demand for natural gas and fuel oil used for heating. Hot summers increase the demand for electricity and the commodities used to generate electricity such as natural gas and coal. A commodity’s price can affect demand. As prices rise, consumers look for ways to use less of the commodity. When gasoline prices rise above a certain level, consumers drive less and purchase more fuel-efficient vehicles. If, as prices increase, there is a major drop in demand, economists say there is high price elasticity. If consumers still need the commodity and don’t have a viable alternative, then demand may stay strong even if prices are high. This would be an example of an inelastic price.
For some commodities, consumers can switch to a different commodity used for the same purposes. If technically possible, consumers will use the commodity with the lowest price. As demand increases for that commodity, then its price increases relative to the other commodity and customers will switch back. Electrical power plants often have the capability to switch between the fuels of coal, oil, and natural gas, depending on price. In pipes and tubing, copper or plastic can be used. Depending on the prices, builders switch between products made from these commodities. Clothing can be made of cotton, plastic synthetic fibers such as acrylic, wool, or a blend of natural and synthetic fibers. Fabric manufacturers can shift depending on the prices of plastics or cotton, assuming customer tastes are accommodated. Soft drinks and processed foods can be formulated with sugar or high-fructose corn syrup.
The ease of switching and any costs involved with modifying the product or process must be considered when determining the price differential needed for substituting one commodity for another. For example, the fuel source (natural gas, fuel oil, or electricity) for residential and commercial buildings is normally fixed based on the heating system installed during construction. The heating system cannot be easily changed when fuel prices vary, so consumers cannot substitute fuel sources. Some processes used to manufacture plastics can use either natural gas or oil as a raw material. These processes can quickly and easily switch back and forth between raw materials depending on commodity prices. Historically, the ratio of the spot price of oil to the spot price of natural gas has been an indicator of when switching will occur. When the ratio becomes large enough it is economical for power generation and industrial uses to switch from using oil to using natural gas. In the long run, companies may redesign their products or processes to reduce the impact of substitution on customer perceptions or costs, as discussed in Chapter 5.
As with supply, government policies and regulations affect demand. The U.S. ethanol market is one influenced by subsidies in the form of tax breaks and tariffs and government mandates in terms of usage. In the United States, the government requires ethanol to be blended into gasoline. In turn, the ethanol market affects the demand for corn.
Technological advances can lead to increases or decreases in demand. For example, in pursuit of energy efficiency, advances in airplane design and manufacturing have increased the use of titanium in aerospace applications. Technology development in nonfood biofuels may lower demand for corn as a raw material for ethanol in the future. If successful in the long run, the demand for hybrid and electric cars is likely to reduce the demand for gasoline, but may increase the demand for electricity. In the long run, this might influence the demand for natural gas to generate electricity.
Changing customer tastes are a major factor affecting demand for some commodities. For example, in China where tea has been traditionally preferred over coffee, from 2014 to 2019, coffee consumption is expected to grow by 18 percent.27 Demand for fiber commodities such as cotton and wool is affected by fashion trends. Food commodities are influenced by changes in dietary habits and fads. For example, lower wheat consumption in the United States has been attributed in part to the popularity of the low-carbohydrate diets in which dieters avoid breads and pastas.28
As was the case for supply, a commodity’s demand is likely to be affected in different ways by a variety of factors, and some factors may increase demand while others may decrease demand. This requires assessing the positive and negative impacts from all relevant factors to forecast the change in demand for the future period of interest. Then, as illustrated with supply, current demand is adjusted upward or downward by the estimated percent change in demand to obtain the future level of demand.
Examining Correlations and Developing Regression Models
So far in this chapter we have described how to analyze and forecast underlying changes in a commodity’s supply and demand. Commodity prices depend on the relationship between supply and demand as perceived by market participants. As supply increases relative to demand, prices tend to decrease. As supply decreases relative to demand, prices tend to increase. However, the market is always changing so when prices do increase, producers have more incentive to produce in larger quantities—and as a result, the gap between supply and demand widens, resulting in decreasing prices. The time frame in which these normal economic cycles occur differs by commodity.
A fundamental analysis uses qualitative judgments of the relationship between supply, demand, and price based on historical patterns to develop a forecast. For some commodities, the quantitative tools based on correlation and regression of historical data can provide useful information for decision making. We will analyze cotton data to demonstrate how to use correlation and regression for developing a long-term price forecast.
Return to the very first step in a fundamental analysis. One method for doing this is using a spreadsheet and plotting annual data for supply, demand, and price on a single-line graph as shown in Figure 4.3. Visually, it appears that as the gap between supply and demand narrows, cotton prices increase, and as the gap widens with more supply than demand, cotton prices decrease.
Let’s create another graph making the relationship between supply, demand, and price easier to examine. In a spreadsheet, this involves subtracting total demand from total supply. This is the supply surplus. The supply surplus and price is plotted on a single-line graph with two vertical axes. For cotton this is shown in Figure 4.4. The graph shows, as would be expected, when the supply surplus increases, cotton prices decline, and when the supply surplus declines, prices increase. Thus for cotton, the supply surplus and price appears to be negatively correlated as would be expected. If the correlation is reasonably strong, a regression equation can be used to forecast the price of cotton.
Figure 4.3 U.S. cotton supply, demand, and price, 1990 to 2015
Source: U.S. Department of Agriculture World Agricultural Supply and Demand Estimates (WASDE) Report on Cotton.29
Figure 4.4 U.S. cotton supply, surplus, and price, 1990 to 2015
Source: U.S. Department of Agriculture WASDE Report on Cotton.30
You can use a spreadsheet or statistical software to calculate a Pearson correlation coefficient (r), which shows the strength of the relationship between two variables. In this example, the variables are supply surplus and cotton price. The value of r is between –1 and 1. An r value of –1 means there is a perfect negative correlation between the variables, so the variables move perfectly in opposite directions to each other. An r value of 0 shows there is no relationship between the variables. When r is 1, the variables will move together perfectly in the same direction. In the example, we used annual data from 1990 to 2015 and the correlation between supply surplus and price for cotton is r = –0.58, which is a relatively strong negative correlation. Therefore, when supply exceeds demand by a greater amount, cotton prices decrease.
It is possible that high correlations between two variables occur just by chance, especially if the number of observations in the sample is low. In the cotton example, the sample size is 26 since the annual data covers the time period from 1990 through 2015. To determine if a regression will be useful for forecasting price is to test the statistical significance of the correlation. To do this test, we should compare the absolute value of r to the critical value of r, published in a statistical table or use a statistical calculator (www.socscistatistics.com). If the absolute value of r is greater than the critical value then the correlation is statistically significant at least p < 0.05, and a simple linear regression model can be developed. The lower the p-value the better. If the value of r is less than the critical value, then you should not proceed to do a simple linear regression.
So, what are the key steps in the correlation significance test? Significance tests assume data are normally distributed. This is likely to be the case if a large enough sample size is used. The most typical threshold of statistical significance is when the probability of the correlation is no greater than 5 percent (0.05). Because, based on logic, we know the expected direction of relationship between supply surplus and price (these should move in opposite directions), use a one-tailed test for significance.
The last piece of information needed to test for statistical significance is the degree of freedom. The degree of freedom is the number of observations in the sample minus two. In our example, it is the number of years observed (n) minus two. For cotton, we have data for 26 years, so the degree of freedom is 26 – 2 = 24. Having more degrees of freedom reduces the critical value of r needed to be statistically significant. If possible, use 15 or more years of supply, demand, and price data. However, this may not be possible if there has been a recent major underlying change in the supply or demand for the commodity.
Using a Pearson correlation coefficient critical-value table for a one-tailed test, with a level of significance of 0.05, and 24 degrees of freedom, the critical value of r is 0.33. For cotton, the absolute value of r (0.57) is greater than the critical value. We conclude the relationship between supply surplus and price is statistically significant, and thus we can use simple linear regression to develop a statistical model to forecast price.
As discussed in Chapter 3, simple linear regression models are easily calculated using an Excel spreadsheet. Create a scatter chart between price and supply surplus. Price must be on the vertical (y) axis because it is the variable you want to predict based on supply surplus. In Excel using the layout tab, click on the data in the scatter chart, then insert a linear trend line and select the option to show the equation and the R2 on the chart. The scatter graph and trend line for cotton are shown in Figure 4.5.
The R2 indicates how well the regression line fits the data. If there is a poor fit, the regression line should not be used as a forecasting tool. As explained in Chapter 3, an R2 value of 1 indicates a perfect fit of the regression line to the data, and all the variance in the data is completely explained by the x-variable. Models with R2 values close to 1 suggest the model will be helpful in forecasting price as long as nothing changes from the past. With real data, R2 values are likely to be lower than 1. In Figure 4.5, for cotton, the R2 of 0.33 shows only 33 percent of the variance in price is explained by the variance in surplus. Thus, although the model can be used to provide guidance it is limited in its predictive power.
Figure 4.5 Simple linear regression: U.S. cotton price and surplus
Source: U.S. Department of Agriculture WASDE Report on Cotton.31
To use the regression model to forecast price, which is the y value, enter your estimate of supply surplus as the x value. The assessment of the underlying factors affecting supply and demand provides the estimated value of supply surplus (x) to use in the equation. Assume you want to forecast the price of cotton in 2020 using the regression model shown in Figure 4.5. In 2015, the U.S. supply of cotton was 16,991 (1,000 bales). Based on the fundamental analysis cotton supply is expected to increase by 2 percent to 17,331 (1,000 bales) in 2020 relative to 2015. Demand in 2015 was 13,900 (1,000 bales), and it is estimated it will increase by 6 percent to 14,734 (1,000 bales) by 2015. Using these estimates, in 2015 the supply surplus is 17,330 – 14,734 = 2,597 (1,000 bales). This is the x value to use in the regression equation (y = –0.0039x + 78.98). Using the regression model, the forecast price of cotton is (y = –0.0039(2,597) + 78.98), which comes to 68.85 cents/lb. Consider this price along with all other information you have and use your judgment to develop a final forecast for cotton in 2020.
Another variable often used to predict price, especially for agricultural commodities, is the stocks-to-use ratio. The stocks-to-use ratio is a percentage, calculated as the ending stock of a commodity in a time period (e.g., a year), divided by its total demand during the time period multiplied by 100. Figure 4.6 shows a scatter chart of cotton price on the y-axis and its stocks-to-use ratio on the x-axis. The correlation between price and the stocks-to-use ratio is r = –0.55. Comparing the absolute value of r (0.55) to the critical value of r (0.39), we conclude the correlation is statistically significant. Thus, a regression model can be developed and used to forecast price based on an estimate of stock to use for cotton.
Some words of caution for using simple linear regression modeling to develop long-term forecasts are in order. Linear regression uses historical data to predict the future. Thus, if there are changes in the fundamental underlying relationships between supply and demand, forecasts based on the regression model will be poor. Look again at Figure 4.1, which shows supply, demand, and price for natural gas. The development of shale gas technology fundamentally changed the supply market beginning in 2005. Thus combining data from before 2005 and after 2005 into a single model would be problematic. The fundamental shift was obvious in the case for natural gas, but for other commodities the changes may be more difficult to identify.
Figure 4.6 U.S. cotton price and stocks-to-use ratio, 1990 to 2015
Source: U.S. Department of Agriculture WASDE Report on Cotton.32
Further, there may be outliers in the data. If you can clearly identify a reason for the outlier, for example perhaps the recession of 2008, you can exclude that data point from the analysis. If not, it would be best to include the data point in the analysis.
For some commodities, you may not find a correlation between price and supply surplus or price and the stocks-to-use ratio. There may be other factors influencing price that can be explored. For example, historically, oil and natural gas prices were correlated, but that is not necessarily the case today. More recently there is a high correlation between corn and oil prices because of the use of corn to make ethanol. If there is a factor highly correlated with price that you can estimate with reasonable accuracy, you can develop a regression equation and use it to help predict price.
Finally, supply and demand tell only part of the story for many commodities. For example, currency strength can affect prices for imported commodities. Price volatility also has been linked to trading activities. In recent years there are concerns that for some commodities, for example oil, prices are not that closely related to underlying supply and demand.33 Thus, the fundamental forecasting process will be unique for each commodity.
Considering Future Prices
The commodity’s futures prices should be examined before finalizing the long-term forecast. Futures contracts, as described in Chapters 5 and 9, are financial instruments in which buyers and sellers agree on the price for delivery of a commodity in the future. Buying a futures contract involves agreeing on a price to take delivery of a fixed amount of a commodity on the date specified by the contract. Sellers of futures contracts agree on a price to deliver a fixed amount of the commodity on the specified date.
Futures contracts are traded only in organized commodity exchanges such as the Chicago Mercantile Exchange (CME) or the London Metal Exchange (LME). There are many commodity exchanges all over the world. Since they are financial instruments, anyone can buy or sell futures contracts as long as they have money to cover a percent of the contract’s value, called a margin. Futures contracts can be bought and sold many times before the delivery date. One confusing thing about futures contracts is the amount of the commodity traded in futures markets has no relationship with the actual amount of the commodity in the physical marketplace.
Futures contracts are standardized in terms of quantity, quality, trading months, and delivery dates. For example, the contract specifications for corn futures as traded on the CME is for 5,000 bushels of no. 2 yellow corn but includes price adjustments for other types of corn and has delivery in March, May, July, September, or December.34 While there are futures available for a large number of agricultural products, metals, and energy products, not all commodities have futures contracts. However, commodities are added to the futures market frequently.
The participants in a futures market are either speculators or hedgers. Speculators are hoping to make money from making the correct decisions about price movements either up or down, and hedgers are hoping to reduce their exposure to price risk. Hedgers are producers or users of the actual commodity. In Chapters 5 and 9 we describe how supply chain managers can use hedging as one strategy to reduce exposure to price risk.
One of the key functions of the futures market is price discovery. Futures prices are not the same as the spot price that is paid for the actual physical commodity. The difference between the spot price and the futures price is called the spread or basis. Normally, the futures price is higher than the spot price because of uncertainty and inventory carrying costs. Occasionally, if the inventory of the physical commodity is unusually low, the futures price may be lower than the spot price.
Thus, futures prices can provide information about the expected trends for spot market prices for some commodities. Futures prices that are closer in time, within the next year to 18 months, tend to be traded at a higher volume and thus provide a better mirror into the price of the physical commodity. If the futures trading volume is low for a commodity, its futures price is not likely to reflect the market’s perceptions about the actual physical commodity’s price.
Further, futures prices are better predictors for some commodities than for others. Arbitrage, in which participants buy or sell in both the futures market and the physical market to profit from price differences, can distort futures prices for some commodities. Thus, determining how much weight to put on the futures prices in forecasting for a commodity requires reviewing the historical relationship between futures and spot prices. If these are closely related, the futures price may be a good predictor of the actual physical commodity’s price.
Developing and Monitoring the Final Forecast
After completing the analysis, judgment is used to develop the forecast from a fundamental analysis. Judgment is needed to identify which supply and demand factors are important and how these factors will change supply and demand in the future. Statistical tools such as correlation and regression can be helpful but are not likely to provide a perfect, error-free forecast. For some commodities, futures prices suggest pricing trends but are not likely to reveal the commodity’s exact price. To develop the final forecast you must weigh everything you have learned about the forecast, the qualitative data, predictions from linear regression models, and the futures price. All this information should be considered in deciding the final forecast. Unfortunately, there is no single magic formula that can be applied across commodities to develop a fundamental forecast. A deep understanding of a commodity’s supply- and-demand markets and their relationships with price is necessary to be successful in long-term forecasting.
With a long-term forecast many things can happen between the time the forecast is developed and when the time period of interest arrives. Supply chain professionals need to continue to gather and monitor supply, demand, and price information after developing a forecast. When new information suggests it is needed, the forecast needs adjustment. Further, if the forecast has changed, managerial decisions based on the initial forecast need to be reviewed and possibly modified.
Summary
In this chapter, we describe an approach for using the market fundamentals of supply and demand to forecast a long-term price. To do a fundamental analysis, a wide range of information needs to be obtained to understand the basics about the commodity’s supply and its demand. This is followed by identifying the key underlying factors affecting supply and demand. Determine which of those factors will affect supply and demand over the period forecasted and the type of change expected. With this information, an assessment of the percent change in supply and demand caused by these factors is determined for estimating the levels of supply and demand in the future.
The statistical tools of correlation and simple linear regression can be helpful in developing a forecast. Supply, demand, and price are plotted on the same line graph to find relationships. Supply surplus and stock-to-use ratios are two variables that may be related to price, but other variables such as stock levels or underlying factors may be more highly correlated with price. If a variable is found to be significantly correlated with price, then using simple linear regression to develop a model to forecast price is appropriate. If the commodity is traded in the futures market, consider the commodity’s futures price and how it may predict the price of the physical commodity. Using judgment and all the information gathered about the commodity, a final forecast is made. The process is iterative in continuing to gather data and monitoring and adjusting the forecast, as well as any managerial decisions based on the forecast, as needed.