North Central Texas Outlook - dfwmaps.com

1 NCTCOG 2040 Forecast May 2015. North Central Texas 2040 Demographic Forecast Over the course of a thirty-five year period, which began in 2005, the North Central Texas area is expected to add nearly 5 million people and over 3 million jobs. The goal of the North Central Texas 2040 Demographic Forecast is to project where these people and jobs are likely to locate within North Central Texas thus allowing sub-regional comparative analysis. For the purposes of this study, North Central Texas is comprised of a 12-county area including Collin, Dallas, Denton, Ellis, Hood, Hunt, Johnson, Kaufman, Parker, Rockwall, Tarrant, and Wise counties. Long-range forecasting is inherently subject to error due to the uncertainty associated with a long time frame and the inability to accurately account for unforeseen and unknowable events.

2 Some of the issues affecting the accuracy of the forecast include: Data errors. While the best available data was used, there could be imperfections in the base-period data due to lack of information or other circumstances. Assumptions. Since it is not possible to know with complete certainly what circumstances will be in the future, some assumptions were necessary. Examples of assumptions included in this forecast are: no major shifts in government policy, no dramatic technological changes, no major changes in human behavior, and no significant changes in the relationships of the various input factors. Geography. In general, as the size of the geography gets smaller, the uncertainty associated with the projection for the area increases. In addition, the known information about the areas varies with some areas having substantial, good data and other areas having much less.

3 The probability that the actual future level of activity will precisely equal the projected level is extremely low for any given area; however, some areas will be closer than others. Modeling. Because the future is unknowable, representations in the form of data models were used to develop a realistic picture of future levels and locations of activity. The practice of data modeling, since it uses abstractions applied by professionals exercising their best judgment, is intrinsically subject to a variety of errors. Geographies There are multiple planning geographies used during the forecasting process. The primary modeling geography is forecast districts. Districts are comprised of smaller areas called Traffic Survey Zones or TSZs. For the 2040 Forecast, the 12 counties were divided into 232 forecast districts and 5,252 TSZs.

4 The TSZs nest to districts, which in turn nest to counties. Digital files of these geographies suitable for use in geographic information systems are available from the Regional Data Center: . They are also available for viewing at: A map showing the geographies and their relationships is provided below. 1. NCTCOG 2040 Forecast May 2015. Forecasting Process The forecasts were developed using a multi-step process. The first step was the development of control totals of households, population, and employment. These were based on data acquired from an independent source1 and served as a benchmark in the forecasting process. The final control totals are presented in the table below. Final Control Totals for the 12-County Area Population in Households Employment Households 2005 2,093,506 5,693,290 3,617,805.

5 2040 3,729,184 10,676,851 6,691,464. 1. Dr. Ray Perryman, The Perryman Group, 2. NCTCOG 2040 Forecast May 2015. District Allocations The second step of the process involved allocating the control totals to forecast districts. The allocation to the districts was accomplished using an NCTCOG adaptation of the Gravity Land Use Model (G-LUM) model2 developed at University of Texas at Austin. The G-LUM system is comprised of two main models. One allocates employment; the other allocates households. These are demand driven models in which employment leads residential location. The models are run on the entire 12-county area and compare each district to every other district. They do not regard city or county boundaries. Key inputs to the models include lagged and base-period employment location by sector, lagged and base-period household locations by income quartile, land use inventories, travel time matrices, and measures of relationships between household types and employment sectors.

6 Data inputs for the models were derived from 2000 and 2010 Census data, NCTCOG major employers and development data, NCTCOG 2005 and 2010 land use inventories, NCTCOG small- area employment estimates; NCTCOG Transportation Department's travel networks, and a variety of other data supplied by local entities. The resulting district-level allocations were made available to local entities for review and comment. This review was conducted in the spring of 2014. NCTCOG reviewed input provided by the local entities and adjusted the district forecasts, where warranted. Special consideration was given to maintain the county-level distributions indicated by the G-LUM model and the figures used in establishment of the regional control totals. Disaggregation to Small Areas The final step in the process was a disaggregation of the district forecasts to TSZs.

7 For this activity, NCTCOG used UPlan, a rule-based urban growth The UPlan model was designed to disaggregate results from other models in this case, forecast district figures out of G-LUM . into smaller areas. Like in the application of the G-LUM models, UPlan was not given any information about jurisdictional boundaries such as cities or counties. UPlan inputs included locally-adopted plans4 and information about locations of existing development such as 2. The model was developed by Dr. Kara Kockelman ( LUM_ ) based on documentation of the DRAM , EMPAL , and LANCON models developed by Dr. Stephen Putman of Putman Associates 3. UPlan was developed by Robert A. Johnston. For more information, see: . 4. This includes future land use, zoning, or other similar plans provided to NCTCOG by local governments or downloaded by NCTCOG staff from city websites.

8 Data collection for this effort was conducted in October of 2013. 3. NCTCOG 2040 Forecast May 2015. residential areas, parks and other amenities, water and floodplains, cemeteries, airports, landfills, and roadways and other transportation features. UPlan used the various inputs along with given population and employment ratios to follow a set of decision rules to determine the likely distribution of the projected changes in households and employment levels between 2005. and 2040. The UPlan outputs of the changes were added to the 2005 base-year data to develop 2040 projections and then aggregated by TSZ. For additional information, see Appendix B. TSZ Local Review In order to help ensure that the small-area projections are consistent with local plans and expectations that might not have been captured elsewhere in the process, one additional step was taken.

9 The TSZ-level figures were made available to local entities for their review and comment. In contrast to the modeling work, local review is a subjective process whereby local entities are given an opportunity to review the model outputs and suggest adjustments. For this, local reviewers use their own expertise, experience, and knowledge along with substantiating information such as approved development plans. An important point to note is that both the objective (modeling) and subjective (local review) processes represent possible outcomes of future development activity and each are subject to the issues associated with long-term forecasting. As the final effort in the process, NCTCOG staff assessed input submitted by local reviewers along with other items that NCTCOG Population The G-LUM model outputs households.

10 Household population6 is estimated separately. The critical factor in the calculation of household population is the average number of persons in each household or household size. Overall, the average household size in North Texas is expected to show an increase over the forecast horizon. However, individual areas districts and TSZs might not show any change or could even indicate a decrease. This was also taken into consideration. Due to household size dynamics, it is possible for an area to experience an increase or decrease in the number of households and yet have an opposite change in household population. Similarly, an area could have no change in the number of households and yet show a change in population. The development of the district-level household sizes included an analysis of the historical household sizes for each district.