FUNCTIONBASEDBIM: APPLIED COMPUTING SCIENCE CATALYZIING PERFORMANCE

Once "systems-thinking" replaces what is currently the mind set or the assumptions held by construction industry practitioners, then top-down, function-based data modeling and analysis will be possible. The aim of this paper is three-fold:

• To provide an overview of how Building Information Modeling (BIM), though energizing the construction industry, is currently limited to just the post-planning and project definition stages. Additionally, BIM's effectiveness is reduced in these latter stages by the fragmented structures and practices of the industry.

• To analyze the effect of adding function-based computational modeling to the current object or geometric-based modeling (BIM) technology, in three key areas: (1) having early, comprehensive, real time planning and budgeting for capital development and building operations, (2) the ability to compute and predict performance standards, and also provide validation and calibration based on actual historical project data, and (3) enable development of integrated, holistic planning, design, procurement, construction and operations management systems that many have envisioned.

• To communicate the implications of the owner, building designer and production team being equipped earlier, more accurately and with multiple options in order to drive the best combination of function, scope, performance and value. This technology enables a performance-based apparatus to drive long term innovation and optimization based on the operations of the completed building. This is instrumental to bringing about the goal of innovative, cost-effective, high performance buildings.

_____________________________________________

Building Information Modeling (BIM) technology has revitalized the construction industry in at least three ways. First, as the most important construction technology of this era, BIM brings the industry into the world of 3D intelligence and cyber discovery. Second, BIM provides the means for greater collaboration, integration and efficiency in everyday design, and also construction, practices. Third, BIM begins to fill what has become a huge innovation hole in construction: in recent decades, construction's declining productivity has often been linked to a lack of innovation, and many welcome BIM as the innovative technology that could improve construction's productivity. Notwithstanding BIM's many merits, its potential is only partially realized.

Most BIM technologies are "object-based." This means that virtual models must be first created (manually) as geometric objects or imported as pre-designed forms — virtual representations of physical spaces and building systems. In the project development process, BIM is typically used only after the major, strategic decisions are made: project purposes and feasibility, planning, site selection, and overall budgeting. This means that BIM’s effectiveness comes into play only in later, esthetic and production stages. However, even at these later stages, BIM’s effectiveness is inhibited by multi-dimensional fragmentation and by the “bottom-up” structures and practices. So, (1) not only is BIM limited to only the later stages of the project development process, but (2) even at these later stages, BIM is restricted and therefore, underutilized. Is there a way to better utilize BIM's productivity and performance potential? If so, what would it look like?

An answer presented in this paper is an emerging generation of BIM that includes, in addition to object-based modeling, function-based modeling. Function-based BIM (or functional modeling) brings “top-down” computational science and data modeling that enables early comprehensive project forecasting, planning and analysis in real-time. The implication is enormous. Not only does it enable, early, real time robust planning, analysis and decision making, but it also enables the industry to overcome its fragmented, linear and “bottom-up” structures and practices. In short, function-based BIM is a key to construction becoming a high performance industry.

There are four purposes to this research and report. The first is to place the concept of function-based computing as integral to construction’s shift from an industrial-era paradigm to a performance paradigm. This shift is essential in building a high performance industry that produces innovative, cost-effective, high performance, sustainable buildings. Function-based BIM technology helps drive the performance paradigm through systems thinking and computational science. Accordingly, the remaining purposes are specific to this technology:

Comprehensive Project Planning — Function-based BIM enables comprehensive, top-down planning based on a project's functions (purposes), scope (magnitude), performance (quality, sustainability and efficiency), and constraints (physical and market). Function-based planning begins well before the first BIM "object" is designed; therefore these "objects" are designed according to the building's function, scope, performance and constraints.

Virtual Project Development Technology — Function-based BIM becomes a critical member of the BIM Family— a technology that enables the transformation of Virtual Design and Construction (VDC) to Virtual Project Development (VPD). The BIM Family becomes the integrated holistic planning, design, procurement, construction and facility operation technology that many have envisioned.

A Standards and Measures Engine —  The establishment of performance standards and measures is foundational to the performance paradigm shift. Function-based BIM provides the computational modeling science, the engine, needed to establish and determine performance standards for the whole-building and also the major building systems. Through such systemic performance standards and measures, the project team can measure and manage the project to substantially higher performance levels.

THE PERFORMANCE PARADIGM  

This is one in a series of white papers presenting a comprehensive solution for construction's productivity and innovation problems. The solution begins with understanding how construction still operates according to an industrial paradigm; the solution ends with an industry-wide shift to the emerging performance paradigm. While the industrial paradigm is characterized by linear logic and fragmentation, the performance paradigm uses computational science and systemic thinking to produce high performance. With new technology systems, building performance is no longer a vague marketing slogan, but a quantifiable quality.    

Building performance is the quality of a building’s operation when measured against a standard. Consequently, without a standard, performance is guesswork. Because standards and measures are currently non-existent for whole buildings, building producers and building processes, no one can know precisely what the industry's performance is, much less how and where to fix it. Why don't standards and measures exist? Construction is massively complex. It has so far been impossible to collect, organize and standardize sufficient building data for a comprehensive analysis. However, new technologies are emerging that use computational modeling to convert construction's complexity into useful data models.         

For the sake of the economy, the environment and energy independence, the industrial paradigm is no longer viable for construction. Conventional buildings tax the environment, and conventional "green buildings" tax the economy. As political, social and marketplace forces begin to demand high performance buildings; the low performance industry must change. Construction must shift in order to perform. 

So, the performance paradigm shift revolves around five key transitions in the industry's structures and practices:  

When these transitions have been made, then the construction industry can be said to have entered the performance paradigm. The benefits of such a shift are many, and include high performance "green" buildings produced by a high performance industry, as well as the national and global recognition — and hopefully emulation — that will attend successful examples of innovation and optimization in such a complex industry.  

COMPREHENSIVE PROJECT PLANNING

Whatever the project type — office building, health care facility, condominium, school, manufacturing facility, parking structure — the nature of building planning, design and delivery will change significantly through the computational science offered by functional modeling. Soon the construction user or delivery team will be able to generate a comprehensive plan that simulates the spatial program, scope requirements, schedule, project (capital) development and operating cost in real-time, before any design form is conceived. This is achieved through the prompted inputs into the functional model of: building function (purposes), scope (magnitude), performance (quality and efficiency), and constraints (physical and market). The simplicity and speed of virtual planning benefit the team by letting it “try out” and examine various scenarios to balance the program, scope, quality and cost. At this point, when the project plan is set, the design can proceed with all the benefits of high-definition information that informs the design, and then analyze it against the plan. Likewise, the procurement, construction, completion and building operation stages are informed by the plan, and then analyzed according to it.

Before illustrating this technology with an early and elementary version of the function-based model, it is helpful to see how the current version, under development, works. This graph below is a high level schematic of the functional modeling technology with associated computational analysis and data structures. It is followed by an outline of the input categories, the tailoring process, and the outputs and reports.

                        

Input of Project Criteria—the user begins by inputting and selecting a few principle criteria into the system, which informs dozens of succeeding default, override-able, and user defined data and information selections. The primary input categories include:  

User Tailoring — The intermediate results are calculated which can be further analyzed and tailored. The user and/or designer proceed to tailor the project at three levels:   

Outputs and Reports - the functional model generates outputs to be something like this:

The structure, accuracy and completeness of the content that populates the databases are critical to the effectiveness of the functional model. This begins with spatial programming data and the relational proportions between the various space uses and the composition of actual room types — number and size of rooms. Some market sectors already have published standards. The illustration below is based on the medical facility market sector, which has good planning and design standards. Examples of this programming expertise include Medical and Dental Space Planning by Jain Malkin, and SpaceMed Guide – A Space Planning Guide for Healthcare Facilities by Cynthia Hayward.

Specialized expertise is also required in building the systems scope and cost data models. Likewise, experienced professionals will access or determine the many adjustment factors that will be applied to the scope and cost data: climate, seismic, market conditions, etc. Initially, the R.S. Means City Cost Index (CCI) may be the base standard — subject to override by experts within respective markets.

In order to handle the complexity of commercial building projects the computational science that makes up the functional BIM system is also complex and extensive. Its development requires significant time, taking computing approaches down many paths until the right combination of data structures and algorithms are discovered.

Statistical analysis is one of the keys to this science. For as complex as the data structures and algorithms are, they are only able to hone in to an 80%+/- accuracy for a particular system or a 90%+/- accuracy for the whole-project. This is not of concern, however, to one who is routinely operating with 100% completed design (contract documents), together with known physical and market conditions. For in such cases — when bidding out the work at the trade level — a variation of 20%+ from the low to high bidder is common. There is no need, therefore, to take the functional modeling precision to something greater than 90%. For it would take substantially more effort, and would suppose an accuracy that is not even possible in the current industry experience — with complete information.

Successive Estimating is a statistical analysis technique proposed by Profs. Steen Lichtenberg of the Technical University of Denmark.¹ This technique yields, for any number of measures (program area, system parameters, cost and other standards of measure): the Mean, Low and High Range Values. These range values can also be tailored based on the desired statements for probability and risk tolerance.

The graph above, entitled, "Statistically Dialing in the functional-based (fBIM) model", illustrates just a sampling of the dozens of possible selections and user-defined inputs that actually exist. The point is the more definition provided the tighter and more precise the results become. When this process is completed, there is a statistical analysis with information ranges and recommended control values for program, scope and cost.

The graph entitled, "Statistically Dialing in the comparative trial analysis", a medical office building includes a group of ophthalmologists, a diagnostic imaging center and an ambulatory surgical center. Once the base trial (A) is created, it can be copied to establish and “try out” other trial scenarios (B and C). This process can continue indefinitely until scenarios are shortlisted for further due diligence and feasibility analysis.

The function-based modeling technology significantly increases the probability of meeting project objectives and parameters early in the project life-cycle. This, in turn, permits the design to progress much more efficiently with less chance of the costly and time-consuming redesign on the one hand, but leaves room for creative design solutions on the other. So the benefits include real time planning at the earliest stages of the project, informed analysis of major solutions and options for high-definition decision making — all at a fraction of the time and cost of traditional methods for planning, design, analysis and management.

VIRTUAL PROJECT DEVELOPMENT

Many understand that integrated Virtual Design and Construction (VDC) is the process and that BIM is the supporting technology. This research and report expands both with Virtual Project Development (VPD) and it’s supporting BIM Family, respectively. By expanding VDC to VPD a project begins with the Virtual Plan which is supported by the “Functional Model” or “Function-based BIM,” a member of the BIM family.

It’s important to note here, that VDC or VPD does not necessarily align with a particular delivery system. That is, while theories about VDC or VPD recognize their ultimate value through higher levels of collaboration and integration between the owner and other project team members, many of the processes and technologies proposed may be successfully applied to traditional project delivery methods, and not just Integrated Project Delivery (IPD). Also, it is anticipated the other delivery systems and organizational structures will emerge as the industry shifts from the industrial-era to the performance paradigm.

The "BIM Family of Technologies Supporting Virtual Project Development" graphic illustrates the relationship between VPD and the BIM Family: the function, geometric, operations and procedure based BIM technologies:

                                         

The function-based modeling technology is presented here as the missing link needed to truly achieve the vision of BIM as the integrated holistic technology. Without functional modeling the project planning process is based on a set of disparate manual and semi-automated processes that significantly limit the project team’s ability to simulate and model various project solutions — both at the project development stage and especially for the completed operating facility and/or infrastructure. However, with functional modeling, a BIM long-term goal is now a near term reality. Through the function-based BIM technology, globally integrated project databases and various applications will enable the team to automate, consolidate and streamline the data and information structures in a way that significantly reduces duplication. This dramatically improves planning, design, execution and, particularly, decision making quality and speed. The following levels represent progressively higher degrees of building and enterprise (i.e. business) planning, programming and analysis according to the above graphic:

fBIM.1 — Building Project Modeling: Independent of or in conjunction with conceptual modeling (gBIM.0), a modeling system will develop a comprehensive plan for the purposes of balancing the spatial program, project quality level, scope of work, first cost budget of the hard costs, for one or multiple trial scenarios.

fBIM.2 — Project Development Modeling: Comprehensive master planning, including all project budget components and even facility-related operating expenses. This is performed before any conceptual drawings are developed based on inputs relating specifically to the purpose of the prospective project.

fBIM.3 — Market to Goal (Objective) Modeling: For the purposes of establishing performance objectives for the project development (and for the operation of the completed building), this BIM level is used to generate “appraised value” comparables (comps) based on current average market conditions, and then industry or sector goals. Initially, this is accomplished by a virtual model only, but as historical information and data is collected from similar projects, the precision of the goals is improved.

fBIM.4 — Enterprise Purpose Modeling: For certain businesses or institutions, multiple enterprise planning scenarios may be analyzed beyond the real estate related operations. This may be based on total revenue or other funding sources, together with operating expenses which include the non-facility as well as the facility related expenses. In this case, BIM becomes “Business Information Modeling” or, better yet EIM for “Enterprise Information Modeling” that would cover the gambit.

GEOMETRIC-BASED BIM:

Geometric-Based Modeling (gBIM) is the technology currently understood as BIM that has grown out of the 3D object oriented and data centric CADD systems used in design and documentation at the whole project level as well as the material assembly levels. The following geometric modeling levels represent progressively greater performance and interoperability among disciplines, systems & trades according to the above VPD graphic:

gBIM.0 — Conceptual Design: This includes planning modeling tools to help facilitate blocking and stacking that becomes the basis for the adjacency and geometric composition of spatial program. The planning model can then be developed with intelligent conceptual architectural systems and assemblies which become effective in collaborating with operational and functional BIM technologies.

gBIM.1 — Design Production: Even though the industry is still in the early stages of building intelligent material and assembly models, BIM has already proven to increase productivity and efficiency in the production of documentation. Many expect design production modeling to become much more performative, with documentation becoming the by-product of that process.

gBIM.2 — Discipline Coordination: This level involves the integration of the architectural, structural and MEP disciplines, for example, including the synchronization of the various trades, clash detection, etc. This yields benefits for the construction trades and vendors by virtue of the documents being better coordinated.

gBIM.3 — Management and Trade Collaboration: The next level serves to develop new standards, outputs, other information and graphical representations which enable greater efficacy and provide more information for the construction trades. This also facilitates the collaboration of the trade professionals in the course of the design.

gBIM.4 — Fabrication/System Collaboration: The development of BIM will increasingly consolidate and eliminate repetitive efforts, especially between the design disciplines and the material or system fabrication disciplines. Ultimately, this may include the actual system, equipment and product manufacturers producing portions of the “contract documents.” Many of the shop drawings are already being done in BIM and, in fact, digital fabrication is being successfully used as well.

The performance paradigm shift will bring robust application and development of the geometric-based BIM technology. The current commodity-based standards and procurement systems effectively prohibit architects and engineers from populating and integrating specific products and systems into its data and model objects. Performance standards, made possible through the function-based BIM technology, will create a great demand for integrating products and systems. Integrated innovation and optimization will draw upon the geometric-based BIM technology to take collaboration, integration, analysis and productivity to new heights.

In a simple example, the diagram, "Geometric-based BIM Illustration, shows how the geometrical model works; in this case, by setting up a building model based on pre-defined structural and enclosure composites, which may be imported into the project model, and generate quantity and cost reports. There has been great progress with the geometrical model, but the industry has still just scratched the surface compared to the potential that exists.

OPERATIONS-BASED BIM

Operations-Based Modeling (oBIM) is the modeling forerunners that are adopted into the BIM Family. These technologies that simulate and automate the aspects of a project that aid in the forecasting, measurement, adjustment and overall operation and management of the completed facility, starting with energy modeling and ending with building automation and facility management systems:

oBIM.1 — Environmental and Energy Modeling: This involves a modeling of the effect of construction and operations of the facility on natural resources. These would include energy, as well as site-related and building-related impact on ecological and environmental systems. These systems are already creating value on many of today’s projects.

oBIM.2 — Building Automation Systems: This involves “the brains” of the actual constructed physical assets that augment and even enable the facility to operate according to the planning and modeling systems. Systems are becoming more intelligent and monitoring/control will likely begin to converge through web services – what is beginning to be referred to as the “Building Information Network” (BIN).

oBIM.3 — Facility Management System: The current FM software applications enable the building operator to manage the facility assets, along with human and other non-facility based assets. It becomes a beneficial database for facility and non-facility assets – from work order management to adds, moves and changes and beyond – performance and measures become integral to the overall operation of the asset.

The growth and development of the sustainability movement is closely tied to the operational modeling development and integration with other non-operational project information and data sets. The concept of TruEx Planning & Budgeting is strategic and can be applied in all of these BIM modules, particularly oBIM. Today’s projects continue to be planned and budgeted based on the capital expenditure (CapEx), which is in conflict with the objectives of sustainability. Unfortunately, the industry doesn't’t yet have readily accessible tools and historical energy and cost data to reorient the budgeting focus. The performance paradigm will be instrumental in filling this void through performance standards, the functional model, the TruEx-based management.

The graph, "Savings Graph with Improved TruEx Driven Design", illustrates the objectives which will be expanded in other papers. Essentially, the first goal is to reduce the CapEx of projects to help fund innovations which support sustainable solutions. The next step is to budget and account for the reduced operating expenses, or the total TruEx on an annual basis. This is described in, “Performance Building” research and report.

PROCEDURE-BASED BIM

INTEROPERABILITY BETWEEN THE VARIOUS BIM MEMBERS

At best, prototyping and populating spatial and system catalogs solutions will satisfy 80% (according to the Pareto Principle) of the incidences on a typical one-off project type. In order to close the gap, the functional model must be very nimble so that the user (design professional) may form and tailor the model so it properly simulates the design vision. This is accomplished in a number of ways, including multiple points of user selections and overrides of default standards together with user defined program, scope and budget entries. As noted above, this is accomplished through statistical analysis which provides ranges which the designer and team may work within.

It is important to address the boundary between technology and art, where technology’s line seems to expand at the expense of art. In fact, the opposite is true, especially on projects where challenging or highly creative design is integral to the project purposes. These technology advancements create the opportunity to yield design efficiency that will allow for more creative energy to be applied. They create an opportunity to yield construction cost effectiveness that will generate savings to help fund the capital costs resulting from the creative design work as well.

Note: The National Institute of Building Sciences has embarked on a significant initiative to develop a series of standards based around BIM, but also coordinated with other industry standards such as Uniformat, MasterFormat and OmniClass by CSI. This is addressed in the National BIM Standards publication. As we move into the functional model vocabulary, starting with naming conventions, (see example below), we find ourselves developing another series of terms and naming conventions that have not yet been developed by these standards organizations. Likewise, within the procurement and construction processes, another vocabulary list has evolved apart from the design. We look to NIBS and CSI to continue to lead such development of these standards and conventions.

     

STANDARDS AND MEASURES ENGINE

The practice of systemic standards and measures is critical to the performance paradigm — for transforming construction into a high performance industry producing innovative, cost-effective, high performance, sustainable buildings. The dilemma is that, without some predictive computational modeling system, it is impossible to establish systemic standards — at the whole building or major systems — for the vast majority of projects. This is the case, whether the measure is gross building area, capital expense, power consumption, labor hours, or any number of measures.

Function-based BIM is a technology that resolves the dilemma. By using accurate historical data from varying control (or comp) projects — real buildings that have been designed and procured, or even built — the functional modeling system is able to simulate such standards of measure as if there were many prior near-identical projects available as “comparables.” These “comps” then act like those used in the real estate appraisal process. It begins by modeling the market average baselines (MAB) for the completed functioning and operating building. It also models various standards associated with the building production process.

The graph, "Statistically Establishing Performance Standards" illustrates the statistical analysis of developing more precise simulation of the market average baseline through the remolding of the appraised values of comparable projects (1, 2 and 3). The market average baseline is used as the control (comp) value in determining the performance objectives.

The “Performance Standards and Measures” white paper presents the psychology and science that attend to this practice. The logic behind performance standards and measures was developed by W. Edwards Deming who taught about the importance of measurement and the importance of a system. In sum, Deming understood an apparently fundamental trait of human behavior: "train people to measure things and they will keep pushing their own standards higher to beat themselves."² In Deming's argument, until people are trained to measure performance, they will not be able to push performance standards higher. When performance standards and measures and function-based computational modeling science are working together according to a system (in this case - starting with the whole building), the result is something like this:

The are striking when looking at a study of labor productivity — the only known production measure the industry has been entreated to. The construction industry has lagged in labor productivity by 20% since 1964, while all other (non-farming) industries have increased by 200% — a 220% difference.³ The productivity potential is quantifiable, and enormous. In this example: Let’s say that the functional model establishes the mean standard value (direct + indirect effort) for a proposed project (design and construction) equal to 120,000 hours. This would be the market average baseline (MAB). If the project team achieved the productivity of the other industries since 1964 that amount could be reduced by 88,300 hours to 37,700 hours (120,000 / (1 + 220% (2.2))). But, realistically, a first project objective may be to increase labor productivity by only 15% (rather than 220%). In such a case, the savings would be more than 15,000 hours (to 105,000 hours) or $750,000 (if the average cost per hour were $50/hr). This would be greater than the combined income that the architect and builder would generate on a traditional project (assuming construction cost of between 10 and 12 million dollars).

Function-based BIM enables the establishment of performance standards: taking real data and information from actual (control or comp) projects and then predicting the standards for the prospective project’s data and information. The effect is this: an estimate (a prediction of the standard) of the prospective project as if there were many — let’s says twenty — prior near-identical control (comp) projects available. The theory and science is this: The real data and information for the twenty actual projects are used to validate and calibrate the functional computational system that is modeling prospective project. If that modeling system is validated and adjusted based on the strength of actual projects, it will be valid and will adjust to establish the standard for the prospective project. This is particularly true if some portion of the actual twenty projects have similar functions (space uses) as the prospective project, even though the scope, location, climate, are significantly different. This is how it works:

After the twenty control (comp) projects are measured, the next step is the calculation of the average (and the variations) of ratios. The ratio numerator is the control project’s actual value and the denominator is the project’s predicted value from the functional model.
4. Finally, the proposed project is processed through the functional model in the same way that the control (comp) projects were. It is essential that this project run through the identical model as the control (comp) projects. The results from the model are then adjusted by the above ratio (control to the proposed project model values). This calculation then produces the standard measure — and also the variation to the standard.

Let’s look at another example, this time using gross building area as the measure: taking the first of the control (comp) projects, the functional model calculates the mean gross building area (GBA) to be 65,000-sf. In this case the actual GBA turns out to be 62,400-sf. The ratio of actual to model for this first project is .96 (62,400/65,000 = .96). This calculation is performed on the other nineteen projects, with the results: Mean = 1.03, 1 Std Deviation = .02. Again, this is the validation and calibration of the functional model by control (comp) projects.

Next, a proposed project is processed through the functional model, and it produces a mean GBA equal to 92,000 sf. Based on the twenty control (comp) projects, the estimated mean GBA — prediction of the standard would be 94,760-sf (92,000 x 1.03 = 94,760) — the market average baseline. A statistical expression of the standard based on one standard deviation: the gross building area projection is between 92,865-sf and 96,566-sf (94,760 +/- .02 x 94,760).

This example shows a relatively tight result from the twenty actual projects. In such an instance, the team can quickly establish the market average baseline (MAB) and proceed to establishing the project objective baseline (POB). In other cases where the standard deviation is much higher, a further analysis into both the control (comp) projects and the accuracy of the information fed into the functional model will be considered.

It is true that such a computing approach in not naturally intuitive. Admittedly, it does appear to be approaching the matter as if coming through the back door. The computational analysis is based on a “top-down” composite of pre-built “bottom -up” space and system objects. To provide further definition or examples would require divulging of proprietary matter. Fortunately, it’s not necessary anyway. The proof is in the pudding since the functional modeling application is validated and calibrated by real life projects.

Other computational science approaches may also develop, but the functional modeling technology is offered here because it works, it is needed anyway for planning and virtual project development, and it is currently the only such system known that serves to establish systemic standards.

CONCLUSION

BIM technology will join performance standards and measures, and, together, they will re-organize the industry around total building performance. Not only will BIM, together with performance standards and measures, be the "silver bullet" for the industry's problems with innovation, they will also play an instrumental role in solving its productivity problems. That the construction industry is responding so positively to BIM shows that innovation is more than welcome.

This expanded generation of BIM — the BIM Family — becomes an integral phenomenon that provides a catalyst to the performance paradigm. This shift is from a low production industry without regard to operating building measures to: a high performance industry producing high performance buildings — functionally, operationally, and environmentally.

 

__________________________________

¹ Professional Construction Management, Barrie and Paulson, 1978, pages 206 to 212

²  Taking the Mystery out of TQM, Capezio and Morehouse, 1993 page 68

³  ENR, July 29 "Productivity Report Calls For Integrated, Efficient Approach," by Bruce Buckley. "Productivity has been a hot-button issue in recent years, particularly following a 2004 analysis by Dr. Paul Teicholz of Stanford University. It suggested that construction labor productivity declined by nearly 20% between 1964 and 2003, while other non-farm industries improved by more than 200%."