Forecasting Payments Made Under Construction Contracts: Payout Curves and Cash Management in the North Carolina Department of Transportation

Abstract

Staff at the North Carolina Department of Transportation (NCDOT) have close control over each highway construction project until the construction contract is let; thereafter, the contractor manages the pace of work, which dictates the flow of payments. These payments make up about one-third of all NCDOT expenditures, so reliable forecasts are important in programming and in cash management. With no other means to predict the pace of construction through the duration of the contract, NCDOT must rely on statistical analyses of past payments to forecast future payments to contractors. Dye Management Group, Inc., which was retained by NCDOT to implement cash management strategies at the department, designed two statistical models of payments to contractors: the first to estimate payments on individual contracts, or the “payout curve,” and the second to estimate total payments made under all contracts in a month. The parameters in both models were initially estimated with data from 4,128 payments made under 336 highway construction contracts completed between August 2000 and June 2002. The first model achieved an adjusted R2 of .93. Although it was useful to engineers in managing individual projects, the model required awkward specifications of seasonal effects to forecast aggregate cash flows. Seasonality was simply accommodated in the second model, which achieved an adjusted R2 of .92. Dye Management Group and NCDOT have operated the second model for more than 2 years and, with a database that has grown to more than 11,000 monthly payments, have consistently achieved mean absolute percentages of error under 10%.