Monday, September 1, 2014

Interest Rates and Supply Chain Management

Interest Rates and Supply Chain Management

To optimize the practice of forecasting, firms must acknowledge the potential for economic or political issues to affect the supply chain process. Generally, any business involved in filling orders for goods and services operates as part of a supply chain. Each link on the supply chain must carry out a specific task in order to deliver products to customers. Because it is consumer oriented, supply chain management is inherently linked to aggregate demand. Changes in economic conditions that create fluctuations in aggregate demand will have a direct impact on supply chain management tasks. For example, if the government increases taxes, there will be a corresponding decrease in net income and aggregate demand, resulting in less production from supply chains. As such, to me it seems that conditions that increase aggregate demand, such as price drops and lower interest rates, should be considered when creating a successful supply-chain forecast model.

Recently, changes in the market for loanable funds market have created uncertainty in the future of aggregate demand. The Federal Reserve has begun putting in motion methods to scale back its Quantitative Easing strategy in response to improvements in the economy. This implies an increase in the federal funds rate, and subsequent rise in interest rates. Rising interest rates imply a fall in net investment and aggregate consumption. Looking forward, from the perspective of supply chain management, it will be relevant to take changing interest rates into account when building forecast models.

In isolation, a forecast model predicting future interest rates can be made using a simple regression model. Specifically, the interest rate of 3-month Treasury Bills (R) can be predicted as a function of the index of industrial production (IP), the rate of growth on the money supply (GM2t = (M2t – M2 t-1/ M2t-1)), and the lagged rate of wholesale price inflation (GPWt = (PWt – PWt-1/ PW t-1)). A multiple regression model can be generated based on these parameters:
Rt = a + b1IP + b2M2t + b3GPWt-1 + e1. Specifically, over the past three months interest rates have wavered between .01 and .04 percent, and forecasts seem to imply that this trend will continue:

More generally, the predicted interest rate can be incorporated in a multiple linear regression forecasting sales of a given product/aggregate demand. However, in order for such a model to prove successful, it must be constantly updated with recent data, as interest rates are particularly prone to sudden changes at any given time. This will help facilitate a continual improvement of the forecasting processes by allowing for adjustment in temporary forecast errors.

In order to conduct a successful forecast and/or causal hypothesis, it is necessary to rely both on forecast models and human intuition. A successful forecaster will both detect a trend and speculate as to why it is happening. However, the future economic climate is heavily dependent on current expectations and media projections. So, I wonder how a supply chain manager can make a successful forecast decision considering several future signals that may contradict each other (i.e. how do they separate the “signal from the noise”). Additionally, multicollinearity is a common problem that forecasters who rely on multiple linear regression analysis seem to run into. How can successful forecasters determine what specific parameters to include in a forecast while avoiding redundancy or including variables that are irrelevant?

Pindyck, Robert, and Daniel Rubinfeld. Econometric Models and Economic Forecasts. Irwin/McGraw-Hill. 1998.
Silver, Nate. The Signal and the Noise: Why so Many Predictions Fail-But some don’t. The Penguin Press. New York. 2012.

No comments:

Post a Comment

Note: Only a member of this blog may post a comment.