The Perryman Group has published annual economic forecasts for almost 40 years and maintains a comprehensive set of national and regional models. We have provided projections throughout the country for a wide variety of clients, including corporations in all major sectors, governmental entities throughout the world, financial institutions, and numerous others with a need for detailed information.
In addition to the standard packages described below, we also produce custom indicator forecasts (sales tax, real estate, energy prices, transportation demand) and small region forecasts (county, city, zip code, or census tract); if you're interested in these or other specialized projections, please contact us .
Key Economic Indicators include nominal and real gross product, nominal and real personal income (by place of residence and by place of work), total employment, wage & salary employment, population, implicit price deflator, consumer price and industrial production indices, labor productivity, nominal and real retail sales, and housing starts.
For the US, the key indicators also include the personal consumption deflator, the producer price index, the 3-month treasury bill and 10-year treasury bond, the nationwide prime rate, and Moody Aaa rating.
The seven largest Texas metropolitan areas include: Austin-Round Rock-Georgetown, Dallas-Plano-Irving, Fort Worth-Arlington-Grapevine, Houston-The Woodlands-Sugar Land, San Antonio-New Braunfels, El Paso, and McAllen-Edinburg-Mission. In addition, the 12 comptroller economic regions are included in the workbook data.
The North American Industry Classification System (NAICS) is a hierarchical classification system of business establishments according to their economic activity. The system consists of 19 industry sectors, listed below with their respective "two-digit" NAICS codes. Each industry sector is comprised of multiple subsectors (with "three-digit" NAICS codes), for a total of 88 subsectors.
Yes. We maintain a comprehensive set of national and regional models and have provided projections throughout the country. These systems include a full range of international, national, regional, state, metropolitan area, county, and area (zip code, census tract, etc.) forecasting and modeling capabilities.
These systems and the accompanying data resources permit forecasting of energy demand and supply, real estate market absorption, retail sales, retail sales taxes, infrastructure needs, industrial performance (manufacturing and service), transportation demand, financial markets, bond feasibility, construction costs, investment returns, and numerous other applications. In addition, projections regarding labor markets, demographics, global markets and trade, and cost factors such as wage rates and capital costs are generated for clients and subscribers on a frequent basis.
Please contact us with more information if you're interested in these or other specialized projections.
The US Multi-Regional Econometric Model was developed by Dr. M. Ray Perryman, President and CEO of The Perryman Group, 40 years ago and has been consistently maintained, expanded, and updated since that time. It is formulated in an internally consistent manner and is designed to permit the integration of relevant global, national, state, and local factors into the projection process. It is the result of more than three decades of continuing research in econometrics, economic theory, statistical methods, and key policy issues and behavioral patterns, as well as intensive, ongoing study of all aspects of the global, US, Texas, and Texas metropolitan area economies. It is extensively used by scores of federal and state governmental entities on an ongoing basis, as well as hundreds of major corporations.
The overall methodology, while certainly not ensuring perfect foresight, has been peer-reviewed on numerous occasions and permits an enormous body of relevant information to impact the economic outlook in a systematic manner.
The US Multi-Regional Econometric Model revolves around a core system which projects output (real and nominal), income (real and nominal), and employment by industry in a simultaneous manner. For purposes of illustration, it is useful to initially consider the employment functions. Essentially, employment within the system is a derived demand relationship obtained from a neo-Classical production function. The expressions are augmented to include dynamic temporal adjustments to changes in relative factor input costs, output and (implicitly) productivity, and technological progress over time. Thus, the typical equation includes output, the relative real cost of labor and capital, dynamic lag structures, and a technological adjustment parameter. The functional form is logarithmic, thus preserving the theoretical consistency with the neo-Classical formulation.
The income segment of the model is divided into wage and non-wage components. The wage equations, like their employment counterparts, are individually estimated at the 3-digit North American Industry Classification System (NAICS) level of aggregation. Hence, income by place of work is measured for approximately 90 production categories. The wage equations measure real compensation, with the form of the variable structure differing between "basic" and "non-basic."
The basic industries, comprised primarily of the various components of Mining, Agriculture, and Manufacturing, are export-oriented, i.e., they bring external dollars into the area and form the core of the economy. The production of these sectors typically flows into national and international markets; hence, the labor markets are influenced by conditions in areas beyond the borders of the particular region. Thus, real (inflation-adjusted) wages in the basic industry are expressed as a function of the corresponding national rates, as well as measures of local labor market conditions (the reciprocal of the unemployment rate), dynamic adjustment parameters, and ongoing trends.
The "non-basic" sectors are somewhat different in nature, as the strength of their labor markets is linked to the health of the local export sectors. Consequently, wages in these industries are related to those in the basic segment of the economy. The relationship also includes the local labor market measures contained in the basic wage equations.
Note that compensation rates in the export or "basic" sectors provide a key element of the interaction of the regional economies with national and international market phenomena, while the "non-basic" or local industries are strongly impacted by area production levels. Given the wage and employment equations, multiplicative identities in each industry provide expressions for total compensation; these totals may then be aggregated to determine aggregate wage and salary income. Simple linkage equations are then estimated for the calculation of personal income by place of work.
The non-labor aspects of personal income are modeled at the regional level using straightforward empirical expressions relating to national performance, dynamic responses, and evolving temporal patterns. In some instances (such as dividends, rents, and others) national variables (for example, interest rates) directly enter the forecasting system. These factors have numerous other implicit linkages into the system resulting from their simultaneous interaction with other phenomena in national and international markets which are explicitly included in various expressions.
The output or gross area product expressions are also developed at the 3-digit NAICS level. Regional output for basic industries is linked to national performance in the relevant industries, local and national production in key related sectors, relative area and national labor costs in the industry, dynamic adjustment parameters, and ongoing changes in industrial interrelationships (driven by technological changes in production processes).
Output in the non-basic sectors is modeled as a function of basic production levels, output in related local support industries (if applicable), dynamic temporal adjustments, and ongoing patterns. The inter-industry linkages are obtained from the input-output (impact assessment) system which is part of the overall integrated modeling structure maintained by The Perryman Group. Note that the dominant component of the econometric system involves the simultaneous estimation and projection of output (real and nominal), income (real and nominal), and employment at a disaggregated industrial level. This process, of necessity, also produces projections of regional price deflators by industry. These values are affected by both national pricing patterns and local cost variations and permit changes in prices to impact other aspects of economic behavior. Income is converted from real to nominal terms using Texas Consumer Price Index, which fluctuates in response to national pricing patterns and unique local phenomena.
Several other components of the model are critical to the forecasting process. The demographic module includes (1) a linkage equation between wage and salary (establishment) employment and household employment, (2) a labor force participation rate function, and (3) a complete population system with endogenous migration. Given household employment, labor force participation (which is a function of economic conditions and evolving patterns of worker preferences), and the working age population, the unemployment rate and level become identities.
The population system uses Census information, fertility rates, and life tables to determine the "natural" changes in population by age group. Migration, the most difficult segment of population dynamics to track, is estimated in relation to relative regional and extra-regional economic conditions over time. Because evolving economic conditions determine migration in the system, population changes are allowed to interact simultaneously with overall economic conditions. Through this process, migration is treated as endogenous to the system, thus allowing population to vary in accordance with relative business performance (particularly employment).
Retail sales is related to income, interest rates, dynamic adjustments, and patterns in consumer behavior on a store group basis. Inflation at the state level relates to national patterns, indicators of relative economic conditions, and ongoing trends. As noted earlier, prices are endogenous to the system.
A final significant segment of the forecasting system relates to real estate absorption and activity. The demand for various types of property is determined by underlying economic and demographic factors, with adjustments to reflect the current status of the pertinent building cycle. In some instances, this portion of the forecast requires integration with the Multi-Regional Industry-Occupation System which is maintained by The Perryman Group.
The overall Texas Econometric Model contains numerous additional specifications, and individual expressions are modified to reflect alternative lag structures, empirical properties of the estimates, simulation requirements, and similar phenomena. Moreover, it is updated on an ongoing basis as new data releases become available. Nonetheless, the above synopsis offers a basic understanding of the overall structure and underlying logic of the system.
The initial phase of the simulation process is the execution of a standard non-linear algorithm for the state system and that of each of the individual sub-areas. The external assumptions are derived from scenarios developed through national and international models and extensive analysis by The Perryman Group. The US model, which follows the basic structure outlined above, was used to some extent in the current analysis to define the demand for domestically produced goods on a per capita basis.
Once the initial simulations are completed, they are merged into a single system with additive constraints and interregional flows. Using information on minimum regional requirements, import needs, export potential, and locations, it becomes possible to balance the various forecasts into a mathematically consistent set of results. This process is, in effect, a disciplining exercise with regard to the individual regional (including metropolitan and rural) systems. By compelling equilibrium across all regions and sectors, the algorithm ensures that the patterns in state activity are reasonable in light of smaller area dynamics and, conversely, that the regional outlooks are within plausible performance levels for the state as a whole.
The iterative simulation process has the additional property of imposing a global convergence criterion across the entire multi-regional system, with balance being achieved simultaneously on both a sectoral and a geographic basis. This approach is particularly critical on non-linear dynamic systems, as independent simulations of individual systems often yield unstable, non-convergent outcomes.
It should be noted that the underlying data for the modeling and simulation process are frequently updated and revised by the various public and private entities compiling them. Whenever those modifications to the database occur, they bring corresponding changes to the structural parameter estimates of the various systems and the solutions to the simulation and forecasting system. The multi-regional version of the Texas Econometric Model is re-estimated and simulated with each such data release, thus providing a constantly evolving and current assessment of state and local business activity.
The process described above is followed to produce an initial set of projections. Through the comprehensive multi-regional modeling and simulation process, a systematic analysis is generated which accounts for both historical patterns in economic performance and inter-relationships and best available information on the future course of pertinent external factors. While the best available techniques and data are employed in this effort, they are not capable of directly capturing "street sense," i.e., the contemporaneous and often non-quantifiable information that can materially affect economic outcomes. In order to provide a comprehensive approach to the prediction of business conditions, it is necessary to compile and assimilate extensive material regarding current events and factors both across the state of Texas and elsewhere.
This critical aspect of the forecasting methodology includes activities such as (1) daily review of hundreds of financial and business publications and electronic information sites; (2) review of all major newspapers in the state on a daily basis; (3) dozens of hours of direct telephone interviews with key business and political leaders in all parts of the state; (4) face-to-face discussions with representatives of major industry groups; and (5) frequent site visits to the various regions of the state. The insights arising from this "fact finding" are analyzed and evaluated for their effects on the likely course of the future activity.
Another vital information resource stems from the firm’s ongoing interaction with key players in the international, domestic, and state economic scenes. Such activities include visiting with corporate groups on a regular basis and being regularly involved in the policy process at all levels. The firm is also an active participant in many major corporate relocations, economic development initiatives, and regulatory proceedings.
Once organized, this information is carefully assessed and, when appropriate, independently verified. The impact on specific communities and sectors that is distinct from what is captured by the econometric system is then factored into the forecast analysis. For example, the opening or closing of a major facility, particularly in a relatively small area, can cause a sudden change in business performance that will not be accounted for by either a modeling system based on historical relationships or expected (primarily national and international) factors.
The final step in the forecasting process is the integration of this material into the results in a logical and mathematically consistent manner. In some instances, this task is accomplished through "constant adjustment factors" which augment relevant equations. In other cases, anticipated changes in industrial structure or regulatory parameters are initially simulated within the context of the Multi-Regional Impact Assessment System to estimate their ultimate effects by sector. Those findings are then factored into the simulation as constant adjustments on a distributed temporal basis. Once this scenario is formulated, the extended system is again balanced across regions and sectors through an iterative simulation algorithm analogous to that described in the preceding section. There are those who maintain that the best forecasts are generated by complex models that capture the interactive forces that drive economic activity. There are others who claim that the optimal approach is to rely on the informed judgment of those who are involved in the process. On this issue, I stand firmly in the middle. I have long held that well-developed models are invaluable tools. They impose logic and consistency on millions of interrelated phenomena and, when properly structured, provide key insights into the ways in which changes in part of the economy work through the entire system. On the other hand, I realize that the knowledge on the streets (both Main and Wall) is equally essential to reliable forecasting. I view my mission for my clients and subscribers as providing the best information I possibly can. I can only do that by combining the two approaches.
As much as some of my colleagues in the quantitative world hate to admit it, there is an irrefutable rationale in statistical theory for using judgmental, non-quantitative information in the preparation of forecasts. Specifically, the desirable property of statistical efficiency (minimum variance) can only be achieved if a prior condition, known as statistical sufficiency, is satisfied. Statistical sufficiency, in turn, requires that all relevant information be used, be it an economic time series published by a government agency or the thoughts and insights of a local building contractor. It’s really pretty simple: the more relevant the information, the better the forecast.
No forecasting technique is perfect. There are no guarantees. Wars, assassinations, natural disasters, technological breakthroughs, and countless other factors can alter the course of the economy in a heartbeat. Subtle changes in the underlying structure of the economy may not be perceptible in the data for decades, and the future policy environment is anything but certain. Consumer and business expectations can shift with the wind, responding to things far removed from local conditions. At The Perryman Group, we don’t promise perfect forecasts. To do so would be patently foolish. We do pledge, however, to use the best information and systems available to provide a reasonable, rational picture of the future course of economic activity. Our expanded modeling systems reflect this commitment which has been consistent and unyielding over the course of the past two decades.
The Perryman Group is a focused team of analysts who know how to address any economic information task and present findings effectively. Our in-house professionals bring expertise in economics, finance, statistics, mathematics, real estate, valuation, systems analysis, engineering, technical communications, and marketing. Dr. Ray Perryman, President and CEO, has 40 years of experience in developing systems, analyzing complex problems, and communicating effectively.