REMI header image

Journal Citings

This article presents the Regional Economic Models, Inc. (REMI) Economic-Demographic Forecasting and Simulation (EDFS) model, which is used for regional forecasting and policy simulation in both the private and public sectors in the United States. The detailed structure of the model is presented. To illustrate the dynamic simulation properties of the model, results of two sample simulations for a REMI multi-area model of a region in Southern California are presented. Post-sample historical forecasts for all U.S. states are provided to evaluate the forecasting capabilities of the model.

A technique for creating regional input-output models based on national input-output technological coefficients is developed and tested. The Regional Purchase Coefficient (RPC), i.e., the proportion of regional demand fulfilled from regional production, is based on substitution between extra-and intra-regional sources in response to relative delivered costs. An RPC estimating equation is fitted using state economic and interstate transportation data. RPCs for all sectors are estimated for use in constructing 500-sector non-survey models for Washington and West Virginia. The A matrices, inverse matrices, and multipliers for suitable aggregations of these models are compared with the corresponding components of recent survey-based models for the same two states. The comparisons indicate that the RPC technique can provide at low cost models that are acceptably accurate for use in regional impact analysis.

Part I of a two-part article discussing regional macroeconomic models as tools for making economic forecasts and simulating the total economic effects of a policy change.

Part II of a two-part article discussing regional macroeconomic models as tools for making economic forecasts and simulating the total economic effects of a policy change.

In this article, the internal structure of the Massachusetts Economic Policy Analysis (MEPA) model is described with special emphasis on the relationships in a state economy that must be considered if the model is to provide adequate answers to inquiries from government officials. Next, selected policies are presented, and their effects are analyzed using the model. Finally, the possibilities and limitations of this approach are summarized.

This paper deals with the problem of modelling sub-county areas that are so small that county level forecasts cannot be used for policy purposes. Starting from a simple specification that relates demand for local goods and services to local and non-local income, a system of equations is developed that can be used as a satellite to a county model to forecast impacts of economic events at the town level.

Five versions of a regional economic forecasting and simulation model are implemented to evaluate the forecasting accuracy and significance for impact analysis of alternative regional labor market closures.

A multiregional policy analysis model should be capable of generating accurate and comprehensive forecasts conditional on alternative values for government policy instruments. While accuracy and comprehensiveness can be accepted as the two multiregional model design goals, the stategies adopted to achieve these goals differ greatly from one model builder to another. In this paper, the strategy is clearly eclectic. It draws on many modeling approaches, including input-output, economic base, neoclassical general equilibrium, Keynesian macromodeling, regional location analysis, segmented labor market analysis and econometric modeling.

Despite the importance of predicting investment expenditures for regional economic forecasting and policy simulation, little has been published on predicting regional investment expenditures. The primary reason is the lack of data on regional investment and capital stocks. Using two constructed investment data sets, this paper specifies and econometrically estimates stock adjustment equations of residential and nonresidential investment for the fifty states plus Washington, D.C. Unique aspects of the approach include maximum use of United States and regional data, and pooled estimation. The estimated pooled equations provide satisfactory historical fits to investment for most states. Also, the paper presents out-of-sample forecasts and simulated investment responses to an exogenous production increase.

In this paper we present a general new economic geography model with multiple industries and regions, full labor and capital mobility, land use in production and consumption, and a dynamic adjustment process in which consumers maximize utility and firms respond to nonzero profits. All industries use intermediate inputs as well as land, labor, and capital. Systems of cities form endogenously within this framework, including asymmetrical urban hierarchies and cities of different sizes and industry compositions. Each urban area has a bid-rent gradient and zones with land uses and densities as in the von Th√ľnen model. The equilibrium depends not only on initial conditions but also on speeds of adjustment. The model is a prototype for empirical implementation, as illustrated with a simulation of the effects of transportation cost reductions.

This paper focuses on the effect of demographic composition on state-level forecasts of labor force participation rates, and on the dynamic response of labor force participation rates to a change in employment opportunity.

In this paper, we describe how Regional Economic Models, Inc. constructs US national and regional forecasting and simulation models, using the US Bureau of Labor Statistics Outlook-2000 forecast. The building procedure extends a traditional input-output model to a dynamic and structural forecasting and simulation model. While much emphasis is on building a consistent US model, we also discuss the linkages and structural differences between the national and regional models. The procedure we present is generally applicable to any regional modelling undertaking. It extracts changing relationships from the national level for use in model building and forecasting at the regional level, where these relationships are not directly observable.

We measure substitution in production for major age-sex groups in ten industries. These estimates are important for productivity studies, for modelling derives demand for labour and for formulating policies that deal with anticipated trends in the age-sex composition of the labour force. We use Sato's two-level CES production to estimate Hicks partial factor price elasticities, with quarterly time-series taken from the Social Security Continuous Work History Sample (1958 to 1975). Our elasticities are generally small and negative but vary considerably across industries. Cross elasticities show complementarity among most groups, except for younger and older females, who are subtitutes.

The Treyz, Friedlaender, Stevens (TFS) regional (i.e., sub-national) modelling approach represents an alternative to constructing regional models using traditional econometric procedures (TEP). In the TFS approach, a model structure based on economic theory is successively calibrated by using information from many sources and parameter estimates at each step from studies that encompass all regions. The use of a maintained structure and of large data sets yields econometric response parameters that are valid for all regions. However, the specific calibration coefficients and the time series for the model of any particular region are based on data for that region only. This makes the behavioural characteristics of the model of one region differ substantially from those of another region but it does not change the basic theoretical structure of the model from region to region. In this article the structural equations used in TFS models are presented. Key structural parameters are identified and their estimates are found in studies across all industries and occupations in time series for forty-eight states.