|
Activity Number:
|
474
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Wednesday, August 1, 2007 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Business and Economics Statistics Section
|
| Abstract - #308961 |
|
Title:
|
A Statistical Model of a Firm's Expenditure
|
|
Author(s):
|
Jonathan Hosking*+ and Bibhas Chakraborty
|
|
Companies:
|
IBM T. J. Watson Research Center and University of Michigan
|
|
Address:
|
, Yorktown Heights, NY, 10598,
|
|
Keywords:
|
business ; finance ; forecasting ; point process
|
|
Abstract:
|
We present a statistical model of the expenditure over time of a firm or other business organization. Assume that the firm "commits" to expenditure by issuing purchase orders for goods or services from outside suppliers. Each purchase order results in one or more payments being made to the supplier: these payment events cause expenditure to be recognized in the firm's accounts. We model expenditure arising from a single commit as a marked point process with parameters that are functions of the attributes of the commit. We model the dates and amounts of future commits, using a regression approach that relates rate of commit to time of year. With these models and a schedule of future releases, we construct forecasts of future expenditure. These can be used by the firm's financial officers to manage expenditure. The models are applied to the capital expenditure of an IBM business unit.
|
