About the Models
Originally known as the Indiana Model of the U.S. (IMUS), IMUS is a medium scale model of the U.S. economy which is used to produce short-run forecasts (at least 12 quarters, with all four quarters of the final year) on a quarterly basis, with monthly updates, and long-run (20-year horizon) projections on a semi-annual schedule. The model currently has 235 variables, 183 of them endogenous. These are determined by 78 estimated behavioral equations and 105 identities.
The structure of the model can be broken down into 6 sectors: equations related to the determination of aggregate expenditures, income, government finance, financial variables, labor market variables and wage/price levels.
Expenditure sector: The model has a Keynesian structure with the level of output determined proximately by aggregate expenditures. In the current version of the model, expenditures are aggregated into 15 categories. Real levels for 3 of these (inventory change, exports of goods and exports of services) and nominal levels for two (federal defense and nondefense purchases) are treated as exogenous. The equations for the real levels of the others are reasonably standard, reflecting variables from the income side of the model, financial sector variables, relative price variables and labor market variables.
Income: The income side of the model has behavioral equations for a number of the components of national income, and also for several variables related to the determination of personal income. Other components of these 2 totals are determined exogenously (for example, rental income and farm proprietors' income) or via identities (for example, employee compensation is given by a wage rate variable times an hours variable). The relationship between GDP and national income is achieved by treating the statistical discrepancy as a residual income category.
Government Finance: The NIPA budget of the federal government is determined by several behavioral equations (for example for transfer payments) and a number of variables which are set exogenously (for example, several tax rate variables). These interact with variables determined elsewhere in the model to determine components of the budget. The state and local NIPA budget situation is similar, but the proportion determined endogenously is higher.
Financial: The most important element of the model's financial sector is a "reaction function" determining the federal funds rate. Changes in payroll employment and in the inflation rate are central independent variables in this equation. Other equations determine the 3-month T-bill rate, the prime rate, the AAA bond rate, M1 and M2.
Labor Market: The central component of the labor market are behavioral equations for the labor force and the unemployment rate. The latter is an Okun's Law type relationship. Given these 2 values, civilian employment is determined by an identity. Other identities determine related employment and man-hours variables.
Wage/Price: The central element of price determination in the model is a Phillips curve type equation which determines the deflator for private nonfarm product. This price variable is a primary determinant of the deflators for the subcomponents of expenditures.
Known as the Indiana Model of Indiana (IMI), this model produces forecasts for the values of employment and income variables for the state of Indiana. Employment is disaggregated into 32 sectors using the North American Industrial Classification System (NAICS). There are 14 manufacturing sectors, 15 non-manufacturing sectors and 3 government sectors. The model currently contains 79 Indiana variables, 5 of which are exogenous. The 74 endogenous variables are determined by 45 behavioral equations and 29 identities.
The forecasting process involves 3 separate steps. The first uses a U.S. Satellite model (USSAT) to forecast employment by industry sector at the national level. The exogenous variables for the USSAT model are provided by the Indiana Model of the United States (IMUS). In the second stage, state level variables are put through a seasonal adjustment process. The final step in the process uses the Indiana Model of Indiana (IMI). This model makes use of the data provided by the USSAT and the seasonal adjustment procedure to create the Indiana forecast.
Short-run (at least 12 quarters, with all four quarters of the final year) forecasts are done on a quarterly schedule and long-run (20-year horizon) projections on a semi-annual schedule.
The Indiana MSA forecast model is used to produce short-run forecasts employment, personal income, and population for 14 Indiana metropolitan statistical areas (MSAs) and the remainder of the state (those counties not in a MSA).
The MSAs included are Indianapolis, Gary, Fort Wayne, Evansville, South Bend, Louisville, Elkhart-Goshen, Lafayette, Bloomington, Terre Haute, Anderson, Muncie, Kokomo and Columbus. (The Evansville and Louisville areas include only the Indiana portions of the MSA.)
The primary data source for the model comes from the U.S. Bureau of Economic Analysis Regional Economic Information System (REIS). REIS data cover employment and income by industrial sector at a county level, The data are annual beginning in 1969, and they are released with a 2-year lag. Population data is from the Bureau of the Census and population projections are from the Indiana Business Research Center. Recent historical data (as well as the forecast itself) are generated using data and forecast results from the Indiana Model of Indiana.
MSA forecasts are produced quarterly.
U.S. and Indiana Projections
In addition to short-run forecasts, CEMR also constructs long-range projections for both the United States and Indiana. The forecasts are produced on a semi-annual basis with a projection horizon of 20 years (80 quarters). There are no separate models for long-range projections—the IMUS and IMI models are simply extended out to the longer horizon.
The first 3 years of the projection utilizes the most recent short-run forecasts from the IMUS and IMI models. Beyond 3 years, a primary assumption driving the projection is that the U.S. economy will achieve and maintain a full employment, steady growth path at some point after the end of the short-run forecast. The length of this transition period depends, in part, on the path forecast in the short run. Consequently, the long-range projection is a "trend" rather than "cyclical" outlook.
For the Indiana economy, in addition to producing a baseline forecast, two alternative projections (a high and low scenario) are produced as well. These alternatives are generated by making moderate adjustments to exogenous variables within the IMI model.
The Indiana Gross State Product (GSP) Projections
In conjunction with long range-projections for state level employment and income generated with the IMI model, a forecast of detailed (2-digit NAICS) gross state product is generated as well. Using data provided by the U.S. Bureau of Economic Analysis, a sector-level econometric model has been developed to provide a measure of value added for selected Indiana industries. When used in conjunction with employment projections, the GSP values provide a measure of productivity by industry within the state of Indiana.
The industries forecast include: wood products, primary metals, fabricated metals, non-electric machinery, electric machinery, autos and parts, other transportation equipment, furniture, plastic products, mining, construction, wholesale trade, retail trade, utilities, communication, transportation, trucking, finance, insurance and real estate, health services and educational services.
As with the IMI long-range projections, the CEMR creates alternative high and low paths for GSP.