You are here

Forecasting Consumer Price Indexes for Food

Located here is an article entitled "Forecasting Consumer Price Indexes for Food: A Demand Model Approach," by Kuo S. Huang. Huang uses an inverse demand function to assess the change in quantity demanded for a variety of goods (including beef, eggs, fruits, vegetables, cereal) based on a one percent change in price of that good. Huang presents a chart that shows, for example, that a one percent increase in the price of poultry would result in a .84 percent decrease in the quantity demanded of poultry. He also gives figures for cross elasticity of demand: A one percent increase in the price of red meat, for example, .91 percent decrease in the quantity demanded of beef.

Huang uses six aggregate food quantities and per capita income to for forecasting consumer price indexes. Huang himself admits that relying on this information may harm the accuracy of his study.

Not only can we not be sure that his information is reliable, the figures he presents are not as useful as I first thought for our model. Most importantly, we are interested in data for very precise demographics It does us little good to see how the American in general responds to a change in price of a particular good (even if this information is accurate): we need to know a middle income Poweshiek County resident who buys 40% of her food locally responds to a change in price. Perhaps Haung's data and methodology will provide us with a starting point for obtaining relevant figures of our own

Comments

Although Huang's forecasting algorithm is interesting, as you already suggest, it is not all that relevant to this project.

It is not that his demographics are not the ones of key interest to us, what he is attempting to forecast is not really related to our system model.

His work focuses on a price - quantity demand relationship, i.e., how does an increase in price generally affect quantity demanded in the aggregated food categories he investigates. This is a much more classic economics forecasting algorithm and much less an agent-based system model.

Although we have already found that fresh, local produce can often cost more than the produce one can buy in the chain supermarket downtown, there are variables at play in the buy fresh and local consumer behaviors that are not considered in Huang's forecasting algorithm.

The idea that price alone affects quantity demanded is just not as relevant when you think in terms of system models, and particularly in the buy fresh, buy local food economies in which we are interested. There are many more variables at play in the system we are modeling.

Our intention too is not to forecast behavior; it is to model observed behavior in order to understand the system. Once we have a good model of how the system operates as it is, the simulation we develop can be used to test some assumptions about introducing change in the system and how it may affect behavior at the system level.