What's a proxy variable?
Some variables are exactly what their label says, and are easily collected and aggregated. Examples include: sales (either in units or value); production (units or value); employment, etc. But for some products or industries these are not easily collected and aggregated.
A proxy variable reflects some aspect of what you're trying to measure, but because of difficulties with data collection or aggregation is not exactly that. Two examples: consumer expenditures instead of product sales; consumer price index instead of product price/cost data.
Below are two examples to illustrate when a proxy variable may be required ...
- How do you define units of clothing? Is a six-pack of socks a unit, or should you count each pair of socks, or each sock?
- How do you account for the large range in value for clothing? A tuxedo or floor-length gown (formalwear) is one unit, but its value (price) will greatly exceed the value of that six-pack of socks. Comparing the two is not meaningful.
- In general, farmers sell their products to distributors or to food processors, not directly to consumers. Food consumers buy from either the distributors (grocery stores usually, which in turn buy from wholesalers, another layer of distribution) or from a combination of food processors (e.g. baked goods and other processed foods) and distributors. Grocery stores now also provide some of the processing (think of pre-prepared foods in the deli, bakery, etc.) There are so many actors and layers in this food chain ... how and when should counting and measuring occur?
- How do you account for food consumed in restaurants? That also originally came from farmers ...