Technical Report: Methodology for Computing Proposed FTA DBE Goals Minnesota Department of Transportation (MnDOT) FY2016-2018 Submitted to: Minnesota Department of Transportation 395 John Ireland Blvd, St. Paul, MN 55155-1899 Submitted by: The Roy Wilkins Center for Human Relations and Social Justice Hubert H. Humphrey School of Public Affairs University of Minnesota July 31, 2015 Copyright © 2015 by the Roy Wilkins Center for Human Relations and Social Justice. All rights reserved. ACKNOWLEDGEMENTS Dr. Samuel L. Myers, Jr., Roy Wilkins Professor of Human Relations and Social Justice and director of the Roy Wilkins Center, Hubert H. Humphrey School of Public Affairs, University of Minnesota, served as the principal investigator for the project, developing the methodology for the analysis and providing overall leadership to the project. Blanca Monter, Associate Director at the Roy Wilkins Center, oversaw the project, and acted as a liaison with the Office of Civil Rights. Dr. Inhyuck Ha, Professor of Economics at Western Carolina University, performed the statistical analysis including the utilization and discrimination analysis. Dr. Ha was assisted by several graduate students at the University of Minnesota. We are grateful to Kim Collins, Director of Civil Rights, Minnesota Department of Transportation and her entire staff, especially Mary Schmidt, Delores Aguirre and Thomas Mebrahtu who compiled and provided the data used to prepare this report. Methodology for Computing MnDOT FTA DBE Goals FY2016-2018 2 TABLE OF CONTENTS ACKNOWLEDGEMENTS .......................................................................................................................... 2 EXECUTIVE SUMMARY ........................................................................................................................... 4 SECTION 1: INTRODUCTION ............................................................................................................. 11 SECTION 2: DATA .................................................................................................................................. 12 SECTION 3: METHODOLOGY TO ESTABLISH OVERALL GOAL ............................................... 12 3.1 DEFINING THE GEOGRAPHICAL MARKETPLACE ............................................................................................................. 12 3.2 UTILIZATION ANALYSIS ...................................................................................................................................................... 13 3.3 DISTRIBUTION OF ESTIMATED EXPENDITURES BY INDUSTRY CLASSIFICATION FOR FY 2016-2018................. 14 3.4 CALCULATING THE WEIGHTED AVAILABILITY MEASURE ............................................................................................. 15 3.4.1 Certified DBE List Method .................................................................................................................................. 15 3.4.2 Dun & Bradstreet Method ................................................................................................................................... 16 3.4.5 The Base Goal .......................................................................................................................................................... 18 3.5 ADJUSTMENT TO THE BASE GOAL ..................................................................................................................................... 19 3.5.1 Data Used for Adjustment to the Base Goal ................................................................................................. 19 3.5.2 Dummy Variable Method .................................................................................................................................... 20 3.5.3 Residual Difference Method ............................................................................................................................... 20 3.5.4 Adjustment ............................................................................................................................................................... 20 3.6 RACE NEUTRAL ANALYSIS ................................................................................................................................................. 21 SECTION 4: THE MNDOT PROPOSED DBE GOALS ...................................................................... 22 SECTION 5: APPENDICES .................................................................................................................... 23 APPENDIX A AVAILABILITY ANALYSIS ........................................................................................................... 23 Table A-1. Weights used for the Availability Analysis ...................................................................................................... 23 PMJ-1......................................................................................................................................................................................................... 25 PMJ-2......................................................................................................................................................................................................... 27 VJM-1 ........................................................................................................................................................................................................ 29 Table A-3. Availability Analysis - D&Bt Method, All NAICS ............................................................................................ 31 PMJ-1......................................................................................................................................................................................................... 31 PMJ-2......................................................................................................................................................................................................... 33 VJM-1 ........................................................................................................................................................................................................ 35 APPENDIX B DISCRIMINATION ANALYSIS...................................................................................................... 37 Table B-1. Description of Variables in the Regression Analysis................................................................................... 37 Table B-2. Summary Statistics for Subcontracts.................................................................................................................. 38 Table B-3. Discrimination Analysis Dummy Interaction Variable Method for Prime Contracts: By Award Amount .................................................................................................................................................................................................... 39 Table B-4. Residual Difference Analysis using Oaxaca Decomposition .................................................................... 40 Methodology for Computing MnDOT FTA DBE Goals FY2016-2018 3 EXECUTIVE SUMMARY This report provides proposed DBE goal and race neutral participation for the Minnesota Department of Transportation (MnDOT) for fiscal years 2016-2018 on Federal Transportation Administration funded expenditures. To satisfy the requirements set forth in the USDOT regulations, availability rates of willing, able and qualified firms must be computed for well-defined geographic market areas. This report was constructed based on the best available information received from MnDOT as well as the governmentpublished secondary data. This report corrects and updates the base availability rate and the adjustment found in the Executive Summary Report of June 18, 2015. The analysis undertaken suggests a proposed agency-wide DBE goal of 7.13 percent for 2016-2018 on FTA-funded projects. This goal was derived in the following manner: ο· ο· ο· A base goal of 5.54 percent was computed. An adjustment to the base goal was made to account for disparities in prime and subcontract awards that cannot be attributed to differences in industry, location, firm size, credit risk, or other characteristics of DBE versus non-DBE contracts. The adjustment was to increase the base goal by 28.79 percent, resulting in the adjusted goal of 7.13 percent (= 5.54 × 1.2879). The maximum portion of the adjusted goal deemed to be achievable by raceneutral means was found to be equal to 36.07 percent. Therefore, the race neutral goal was computed to be equal to 2.57 percent (= 7.13 × 0.3607) and the raceconscious goal was computed to be equal to 4.56 percent (= 7.13 × (1 – 0.3607) = 7.13 × 0.6393). Table 1 provides the detailed breakdowns: Table 1. Proposed MnDOT FTA DBE Goal FY2016-2018 Goals as June 18 Base Goal 5.71% Discrimination Gap for Adjustment 15.02% 6.57% Adjusted Goal Race-Neutral Portion 29.65% 1.95% Race-Neutral Goal Race-Conscious Portion 70.35% 4.62% Race-Conscious Goal Goals as July 21 5.54% 28.79% 7.13% 36.07% 2.57% 63.93% 4.56% Source: Author's calculation using the FTA data files. Methodology for Computing MnDOT FTA DBE Goals FY2016-2018 4 BACKGROUND As a recipient of federal transportation dollars awarded through the U.S. Department of Transportation’s Federal Transportation Administration (FTA), MnDOT is required to establish and submit a three-year goal to the FTA for review. This goal is to be established in compliance with the federal regulations governing the Participation by Disadvantaged Business Enterprises in Department of Transportation Financial Assistance Programs (hereafter referred to as “USDOT regulations”). The USDOT regulations instruct state and local grant recipients to follow a two-step process to establish their annual DBE goal [49 C.F.R. §26.45]. The analysis conducted by the Roy Wilkins Center complies with these guidelines. METHODOLOGY In order for the MnDOT FTA DBE goal to satisfy the requirements set forth in the USDOT regulations, availability rates of willing, able and qualified firms must be computed for well-defined geographic market areas. We established three different geographic market areas. Two were based on political jurisdictions and the other one was based on zip codes where there were substantial numbers of prime and subcontractors for MnDOT contracts between fiscal year 2012 and 2014. Relevant industries with MnDOT contracts were identified by examining the distribution of MnDOT contract dollars by industry classification for contracts awarded between October 1, 2011 and September 30, 2014. We then estimated the distribution of anticipated MnDOT contract dollars by industry classification for FY 2016-2018 from the information about future projects that was provided by MnDOT. Availability rates were computed from multiple data sets and were appropriately weighted to produce a base goal. The base goal was then adjusted to account for disparities in prime contract and subcontract award amounts. The result is the proposed goal. The proposed goal was further partitioned between a race conscious and race neutral portion using a methodology upheld by the 3rd Circuit Federal Court in GEOD v. New Jersey Transit and published in the peer reviewed journal, Applied Economics Letters. DATA The research team obtained all prime contract and subcontract files from the Office of Civil Rights, Minnesota Department of Transportation for the years 2012 to 2014. Obvious data entry errors, improbable measures, possible duplicates and related questionable items were flagged and forwarded to MnDOT staff for clarification and/or correction. Contract files were supplemented with data obtained from Dun & Bradstreet. Other data used included: the MnDOT DBE Certification Directory and the U.S. Census ZIP code Business Patterns (ZBP) data for 2012. Methodology for Computing MnDOT FTA DBE Goals FY2016-2018 5 UTILIZATION The utilization analysis shows that 100 percent of prime contract dollars were awarded to non-DBE prime contractors while 1.8 percent of contract dollars were awarded to DBE subcontractors. The vast majority of prime contract dollars were awarded to Minnesota firms (95.9%). The bulk of contracts and contract dollars awarded were in the professional and technical service industry. Table 2. DBE Share of FTA Contract Amount N Contract Amount Percent Total 50 $8,613,712.90 DBE 0 $0.00 0.00% 50 $8,613,712.90 100.00% Total 3 $154,223.34 Sub Contracts DBE 3 $154,223.34 100.00% 0 $0.00 0.00% 3 $154,223.34 1.79% Prime Contracts Non-DBE Non-DBE Both Prime and Sub Contracts DBE Source: MnDOT FTA contracts: Oct. 1, 2011 - Sep. 30, 2014 GEOGRAPHIC MARKET AREAS In order for the MnDOT FTA DBE goal to satisfy the requirements set forth in the USDOT regulations as well as comply with the Supreme Court’s narrowly-tailored standard, the DBE goal must be based on a narrowly-defined geographic market. In order to define the geographic market in such a manner that the vast majority of contract dollars awarded would be incorporated in the definition, the research team considered two broad methods: a) political jurisdiction method (PJM), based on the counties where contracts were awarded; and b) virtual jurisdiction method (VJM), based on the zip code location of contracts and/or contractors in the client’s database. All methods yield different counts or estimates of the numbers of firms within industry codes, and accordingly will yield alternative measures of availability. For this report, the following alternative markets were used. Methodology for Computing MnDOT FTA DBE Goals FY2016-2018 6 Table 3. MnDOT Geographic Market Areas (GMAs) for FTA DBE Goals GMA Definition Total Amount Share of Dollars Political Jurisdiction Method (PJM) PJM-1 All Minnesota counties $8,405,636.07 95.9% PJM-2 Ranked Counties in USA where the total contract dollars for the sum of the counties exceeds 75% and the marginal contribution to the overall total is at least 1% $8,596,667.24 98.0% Ranked zip codes anywhere in the USA where the total contract dollars awarded for the sum of the zip codes exceeds 75% and the marginal contribution to the overall total is at least 0.25% $8,690,942.24 99.1% Total Contract Amount (Primes and Subs) $8,767,936.24 Virtual Jurisdiction Method (VJM) VJM-1 Source: FTA contracts between Oct. 1, 2011 and Sept. 30, 2014 The first method, PJM-1 represents the State of Minnesota. The second method, PJM-2, defines those counties where there are enough contracts to represent the Minnesota counties where the total contract amount exceeds 75 percent and the marginal contribution of each county to the overall total contract amount is at least 1 percent. The other measure, VJM-1, is similar to PJM-2, but focuses on zip codes. The distribution of dollars per GMA is presented in Table 3 as well. Each geographic market area has more than 95 percent of contract dollars. DISCUSSION OF AVAILABILITY METHODS The research team obtained from MnDOT a list of all firms that were in their various databases, which include prime contractors and subcontractors, certified DBEs. NAICS codes for the firms were obtained from the BizTrak database, and Dun & Bradstreet (D&B). When no NAICS could be found from D&B, observations were not used in the weighted availability counts. The research team also obtained from MnDOT the list of projects that MnDOT expects to undertake and identified 54 separate six-digit NAICS codes associated with comparable Methodology for Computing MnDOT FTA DBE Goals FY2016-2018 7 projects from the period of 2011-2014. Weights were obtained by using the projected expenditures provided by MnDOT.1 Dun and Bradstreet Method The research team obtained access to Hoover’s database, a subsidiary of Dun & Bradstreet, to compute the number of firms in each relevant NAICS code within a specified geographic market area. This research product covers some 900 industries, which include about 65 million corporations and other entities (i.e. government agencies, partnerships, non-profits, and educational institutions) and 85 million individuals. For the state of Minnesota, information included information on headquarters, branches, and single locations.2 The availability rate is computed by finding the weighted share of women- and minorityowned businesses within each NAICS code for a specified geographic market area. Unlike the other methods, the D&B method uses “self-reported” minority/gender designations. Thus, the D&B method can include firms that are not MnDOT certified DBEs. On the other hand, not every certified DBE is included in this database because a requirement of inclusion is the existence of a DUNS number. According to Hoover’s customer service, D&B contacts firms directly to verify their gender or minority status and checks with third party sources and proprietary databases for further verification. DBE List Method We obtained the list of certified DBEs from MnDOT. The numerator in the availability ratio is the number of certified DBE firms for specified NAICS codes within a given geographic market area. The denominator is the number of firms in the County Business Patterns (CBP) or ZIP code Business Patterns (ZBP) database, depending on the definition of the geographic market area, for specified NAICS codes within a geographic market area. The numerator and denominator come from different sources. The numerator counts firms and the denominator counts establishments with paid employees. The Census Bureau explains: An establishment is a single physical location at which business is conducted or services or industrial operations are performed. An establishment is not necessarily equivalent to a company or enterprise, which may consist of one or more establishments. A single-unit company owns or operates only one establishment. A multi-unit company owns or operates two or more establishments. http://www.census.gov/econ/cbp/methodology.htm, accessed, June 1, 2015 3 (Census Bureau, County Business Patters, “How the Data Are Collected (Coverage and Methodology)”). According to Department of Transportation regulations, the availability rate should be weighted by the “amount of money to be spent” in each industry. The research team requested a copy of MnDOT’s estimated expenditures for the next three years, broken down by NAICS code. Projected expenditures for the next three years included in the State Transportation Improvement Program (STIP) 2016-2018 data files were provided and categorized by type of work. In order to calculate the weights for the availability analysis, the research team categorized projected expenditures by NAICS code. The result was 42 NAICS codes. 2 Headquarter: indicates that the company has subsidiaries or branches; branch indicates a company location other than the headquarters; and single location indicates that the company does not have any subsidiaries or branches. 3 As a result, the denominator includes many establishments that might not normally contract with MnDOT. 1 Methodology for Computing MnDOT FTA DBE Goals FY2016-2018 8 THE BASE GOAL Depending on the method used to calculate availability rates, each geographic market area captures a different share of available contract dollars. As a result, each method also yields a different DBE availability goal for each geographic market. In order to derive a single base goal for MnDOT that is based on all the goals calculated for each geographical market, it is necessary to weight each geographical market-specific goal according to the percentage of contract dollars awarded in that area. The two different methods use data that report multiple industries for many of the firms in their databases. Table 4 presents the details of the weighted availability rate using all NAICS codes level. This base goal is used in subsequent analyses. Table 4. FTA Availability Rates and Base Goal Method DBE List Method PJM-1 0.83% PJM-2 1.69% VJM-1 Base Goal 0.02% 5.54% D&B Method 9.16% 12.21% 9.31% Source: Author's calculation using FTA data files, DBE, D&B, and US Census data ADJUSTMENTS TO THE BASE GOAL According to the USDOT regulations, recipients of federal funds are required to adjust their base goal in light of other evidence regarding the market area [49 C.F.R. §26.45(d)]. One valid reason for adjusting the goal would be if the analysis showed discrimination, either in the overall market place or in the specific agency or governmental unit undertaking the procurement and contracting process. A standard method for estimating discrimination is to compute the difference in contract award amounts attributable to DBE status. This requires enough observations on DBE awards to obtain regression estimates. Because of the unique nature of the FTA contracts, it was not possible to estimate such standard regression models or separate regressions for DBEs and non-DBEs using FTA data alone. There were no DBE primes in the dataset. All of the non-DBEs were primes. There were no subcontracts that were nonDBEs. Moreover, many of the observations were individuals on whom we had no information on industry classification or other relevant factors. To remedy this problem, we identified the NAICs codes for FTA prime contracts and subcontracts. We merged the FTA contract files with the FHWA subcontract awards only for this restricted set of NAICs codes. Methodology for Computing MnDOT FTA DBE Goals FY2016-2018 9 We identified five plausible measures for the adjustment and averaged across them using three different statistical methods. The first computes the unadjusted difference in award amounts between non-DBEs and DBEs arising from the FTA contracts. The second estimates a log-linear regression model and controls for FTA, DBE and an interaction term of FTA and DBE contracts. This interaction term permits us to estimate the difference in contract award amounts due to DBE and FTA awards. The final method decomposes the gap into explained and unexplained portions and computes the share of the unexplained portion that is due to FTA awards. The average across these methods yields an adjustment of 28.79 percent. RACE NEUTRAL ANALYSIS In compliance with federal regulations, state and local transit authorities must identify the maximum feasible portion of the DBE goal that can be achieved through race-neutral measures and the percentage of the goal that can only be achieved through race-conscious measures [49 C.F.R. §26.51(a)]. Myers and Ha have pioneered the use of a detailed econometric procedure that maximizes the race-neutral component of the DBE goals.4 This method has established a rigorous standard for maximizing the race neutral portion of the overall DBE goal.5 The logic of the analysis is that some share of DBE dollars awarded would have gone to DBEs without goals. One can compute the share of dollars that would have gone to DBEs without goals for contracts and firms that are comparable. This method requires the estimation of a regression model that controls for a host of relevant variables. The race neutral analysis uses the best regression model for predicting DBE contract amounts with and without goals. The data set is the merged data set combining FHWA and FTA contracts with the same NAIC codes found in the FTA dataset. Following this specification, our analysis shows that 36.07 percent of the goal can be achieved through race neutral measures and 63.93 percent can be achieved through race conscious goals. Myers, Samuel L. and Inhyuck “Steve” Ha. "Estimation of Race Neutral Goals in Public Procurement and Contracting," Applied Economics Letters, 2009, vol. 16, issue 3, pages 251-256. 5 2010-10-19, Civil Action No. 04-2425, GEOD CORPORATION, et al., Plaintiffs v. NEW JERSEY TRANSIT CORPORATION, et al., Defendants. 4 Methodology for Computing MnDOT FTA DBE Goals FY2016-2018 10 SECTION 1: INTRODUCTION Minnesota Department of Transportation (MnDOT) is a recipient of federal transportation dollars awarded through the U.S. Department of Transportation’s Federal Transit Administration (FTA). As such, MnDOT is required to establish a three-year goal and submit it to the FTA for review (Final Rule, 2010; FTA Memorandum HCR-1, March 16, 2010). This goal must be in compliance with the federal regulations governing the Participation by Disadvantaged Business Enterprises in Department of Transportation Financial Assistance Programs (hereafter referred to as “USDOT regulations”). The USDOT regulations instruct state and local grant recipients to follow a two-step process to establish their annual DBE goal [49 C.F.R. §26.45]. Step 1: Determine the base figure for the relative availability of DBEs in the specific industries and geographical market from which DBE and non-DBE contractors are drawn. Step 2: Adjust the base figure in light of other evidence regarding the market area. This two-step process is designed to satisfy several important objectives [49 C.F.R. §26.1]. According to USDOT regulations, DBE programs are designed to: (a) Ensure non-discrimination in the award and administration of DOT-assisted contracts in the Department’s highway, transit, and airport financial assistance programs. (b) Create a level playing field on which DBEs can compete fairly for DOT-assisted contracts. (c) Ensure that the Department’s DBE program is narrowly tailored in accordance with applicable law. (d) Ensure that only firms that fully meet this part’s eligibility standards are permitted to participate as DBEs. (e) Help remove barriers to the participation of DBEs in DOT-assisted contracts. (f) Assist the development of firms that can compete successfully in the marketplace outside the DBE program. (g) Provide appropriate flexibility to recipients of federal financial assistance in establishing and providing opportunities for DBEs (emphasis added). The MnDOT DBE program and goal-setting process is designed to meet these objectives. On behalf of MnDOT, the University of Minnesota research team completed the required Methodology for Computing MnDOT FTA DBE Goals FY2016-2018 11 two-step process outlined in the USDOT regulations. This report details the methodology used to provide MnDOT with the information it needs to make its decision about the setting of the DBE goals for 2016-2018. SECTION 2: DATA Staff at the Office of Civil Rights, Minnesota Department of Transportation, provided the research team with electronic copies of all known prime contract and subcontract files for the years 2011 to 2014. Obvious data entry errors, improbable measures, possible duplicates and related questionable items were flagged and forwarded to MnDOT staff for clarification or correction. Contract files were supplemented with data obtained from Dun & Bradstreet. Other data used included: MnDOT DBE Certification Directory, MnDOT Vendors List, and the U.S. Census Zip code Business Patterns (ZBP) data set for 2012. SECTION 3: METHODOLOGY TO ESTABLISH OVERALL GOAL The USDOT suggests rather than prescribes how the base figure of relative availability of DBEs is to be calculated [49 C.F.R. § 26.45(c)]. The base figure for the MnDOT goal was calculated using several variations of the availability methodology proposed in the USDOT regulations. The methodologies performed and results produced are summarized here. 3.1 Defining the Geographical Marketplace In order for the MnDOT FTA DBE goal to satisfy the requirements set forth in the USDOT regulations as well as comply with the Supreme Court’s narrowly-tailored standard, the DBE goal must be based on a narrowly-defined geographic market. The Wilkins Center analyzed contracts awarded between October 1, 2011 and September 30, 2014. Our analysis of contracts confirmed that the majority of contracts (92.6 percent for prime contracts) and contract dollars (97.6 percent for prime contracts) were awarded to Minnesota-based firms.6 A small portion of contract dollars was awarded to firms located in Canada (1.7%) and Texas (0.7%). Table 5. Prime Contracts by State Contracts Percent Contract Amount Canada 1 2.0% $142,427.83 Minnesota 46 92.0% $8,405,636.07 North Dakota 2 4.0% $4,999.00 Texas 1 2.0% $60,650.00 Total 50 100.0% $8,613,712.90 Source: MnDOT FTA contracts: Oct. 1, 2011 - Sep. 30, 2014 6 Percent 1.7% 97.6% 0.1% 0.7% 100.0% Subcontracts are not reported at the state level due to the small number of contracts. Methodology for Computing MnDOT FTA DBE Goals FY2016-2018 12 In order to define the geographic market in such a manner that the vast majority of contract dollars awarded would be incorporated in the definition, the research team considered two broad methods: a) political jurisdictional method (PJM), based on the counties where contracts were awarded; and b) virtual jurisdictional method (VJM), based on the zip code location of contracts and/or contractors in the client’s database. All methods yield different counts or estimates of the numbers of firms within industry codes, and accordingly will yield alternative measures of availability. For this report, the following alternative markets were used. As shown in Table 6 below, the first method, PJM-1, represents the State of Minnesota. The second method, PJM-2, defines those counties where there are enough contracts to represent the Minnesota counties where the total contract amount exceeds 75 percent and the marginal contribution of each county to the overall total contract amount is at least 1 percent. The other measure, VJM-1, is similar to PJM-2, but focuses on zip codes. Table 6. Geographic Market Areas (GMAs) GMA Definition Political Jurisdiction Method (PJM) PJM-1 All Minnesota counties PJM-2 Ranked Counties in USA where the total contract dollars for the sum of the counties exceeds 75% and the marginal contribution to the overall total is at least 1% Virtual Jurisdiction Method (VJM) VJM-1 Ranked zip codes anywhere in the USA where the total contract dollars awarded for the sum of the zip codes exceeds 75% and the marginal contribution to the overall total is at least 0.25% Source: FTA contracts between Oct. 1, 2011 and Sept. 30, 2014 3.2 Utilization Analysis As shown in Table 2 above, the utilization analysis shows that 100 percent of prime contract dollars were awarded to non- DBE prime contractors while 1.8 percent of contract dollars were awarded to DBE subcontractors. The vast majority of prime contract dollars were awarded to Minnesota firms (97.6%). The bulk of contracts and contract dollars awarded were in the professional and technical service industry. Table 7 shows that 60 percent of prime contracts were awarded to private individuals. Methodology for Computing MnDOT FTA DBE Goals FY2016-2018 13 Table 7. Prime Contracts by Type of Entity Contracts Percent Contract Amount 1 2.0% $1,182,456.60 13.7% Private Company 19 38.0% $7,352,772.30 85.4% Private Individual 30 60.0% $78,484.00 0.9% Total 50 100.0% $8,613,712.90 Source: MnDOT FTA contracts: Oct. 1, 2011 - Sep. 30, 2014 100.0% Not-for-Profit Agency Percent 3.3 Distribution of Estimated Expenditures by Industry Classification for FY 20162018 According to the Department of Transportation regulations, the availability rate should be weighted by the “amount of money to be spent” in each industry. The research team requested a copy of MnDOT’s estimated expenditures for the next three years, broken down by NAICS code. Projected expenditures for the next three years included in the State Transportation Improvement Program (STIP) 2015-2018 data files were provided. Projected expenditures were categorized by type of work. However, in order to calculate the weights for the availability analysis, projected expenditures must be categorized by NAICS code. In collaboration with construction engineers in MnDOT, NAICS codes have been assigned to each of the relevant projects listed in the STIP data files. Due to the important of the expenditure projection in calculation of the weights, NAICS codes from the past contract file is assigned to each work type based on the description of the work type and the NAICS definition. There were 93 work types with MnDOT’s expenditure projection in the STIP report, with 43 distinct 6-digit NAICS, however, in the FTA contract file, there are 13 available distinct NAICS codes where there were only one distinct NAICS code in common between the two files, which makes it hard to use the conventional method of calculating weights. As an alternative, the research team used NAICS codes from both STIP projection data and FTA contract file and calculated the weights based on summed dollar amount of past- and projected-expenditures. Step 1: Obtained the projected expenditures for the next three years (STIP data) Step 2: Identified six-digit NAICS codes for each projected contract based on its type of work Step 3: Combined expenditure amount between STIP data and FTA contract data by the six-digit NAICs. The weights by NAICS code are shown in Table A-1 in Appendix A. Methodology for Computing MnDOT FTA DBE Goals FY2016-2018 14 3.4 Calculating the Weighted Availability Measure The formula below, which was prescribed in the Final Rule (1999), was used to perform the availability analysis. This formula computes the weighted share of DBE firms within the geographic market area. The weights are the share of dollars DBEs can expect to be awarded within each industrial NAICS code. π΄π£πππππππππ‘π¦ π ππ‘π ππ’ππππ ππ π·π΅πΈπ ππ ππ΄πΌπΆπ1 = [( ) × π€πππβπ‘1 ππ’ππππ ππ πππ‘ππ πΉππππ ππ ππ΄πΌπΆπ1 ππ’ππππ ππ π·π΅πΈπ ππ ππ΄πΌπΆπ2 +( ) × π€πππβπ‘2 + β― ] ππ’ππππ ππ πππ‘ππ πΉππππ ππ ππ΄πΌπΆπ2 To calculate the availability rates, the number of ready, willing, and able minority and women-owned firms in the defined geographic market area were compared to the total number of ready, willing, and able firms in the same geographic market area. Although the U.S. Department of Transportation does not dictate a single method for calculating the DBE availability rate or the source of the base goal, it does suggest several methods for calculating the base goal. Moreover, the federal rule mandates that all available data be used. In order to produce the best estimate of the available number of women and minority-owned firms in MnDOT’s geographic market areas, the research team used a combination of relevant methods suggested in the regulations. The U.S. Department of Transportation recommends that grant recipients use 1) a DBE directory and census data, 2) a bidders list, 3) data from a disparity study, 4) the goal of another DOT recipient, or 5) alternative methods that are “rationally related to the relative availability of DBEs in your market” [49 CFR §26.45(c)]. The availability rates from the last statewide disparity study were deemed too old. The availability rates from the most recent local disparity studies (e.g., Minneapolis) were deemed inapplicable. Accordingly, the research team adopted two different measures of availability, each of which meets the standards articulated in the federal regulations. These two methods use alternative data sources, each with different advantages and disadvantages. 1) MnDOT Certified DBE Directory 2) Dun & Bradstreet database 3.4.1 Certified DBE List Method Step 1: The certified DBE directory includes the firms that are certified as DBEs by MnDOT. The research team obtained the list of DBEs. The directory contains detailed information on each DBE firm including NAICS codes. Step 2: Ensured the consistency of the NAICS standards used by converting the original NAICS codes into NAICS 2012 codes if needed. Methodology for Computing MnDOT FTA DBE Goals FY2016-2018 15 Step 3: Obtained the 2012 ZIP Code Business Patterns (ZBP) and County Business Patterns (CBP) from the Census website (most recent year available to prepare the availability ratios). This dataset provides, for each NAICS code, the total number of firms headquartered in a given zip code. This list is used for the denominator in the availability calculation, while the merged DBE list is used to determine the numerator.7 Step 4: Narrowly defined the geographic markets as PJM–1, PJM–2, and VJM–1. Step 5: Transformed the list into the database where the NAICS code of each firm is the unit of observation, in accordance with the NAICS code-based definition of the prescribed availability formula. Step 6: Derived, in each geographic market and for each relevant NAICS code, the total number of DBE firms (using the DBE list) and the total number of firms (using the ZBP or CBP list). Step 7: Calculated the share, or the un-weighted availability, of DBEs, which is the ratio of DBEs to total firms in a given relevant industry (NAICS code) and narrowly tailored geographic market. Step 8: In some cases the NAICS code for future expenditures produces zero counts of total available firms. The resulting DBE share is not calculable and setting the value to zero provides the erroneous conclusion that there are no available DBEs. Therefore, these cases were treated as missing values, and the values were replaced by the un-weighted mean of the values with positive denominators. Step 9: Calculated, for each geographical area, the weighted availability rate of DBEs according to the prescribed availability formula. Step 10: Averaged availability results with and without imputing means in order to obtain the final availability rate for each geographic market area. Please see Tables A-2 in Appendix A for the availability rates for three different geographic market areas. 3.4.2 Dun & Bradstreet Method Step 1: The research team obtained access to Hoover’s database, a subsidiary of Dun & Bradstreet (D&B), to compute the number of firms in each relevant 7 The Census Bureau explains: An establishment is a single physical location at which business is conducted or services or industrial operations are performed. An establishment is not necessarily equivalent to a company or enterprise, which may consist of one or more establishments. A single-unit company owns or operates only one establishment. A multi-unit company owns or operates two or more establishments. (Census Bureau, County Business Patters, “How the Data Are Collected http://www.census.gov/econ/cbp/methodology.htm, accessed, June 5, 2012). As a result, the denominator includes many establishments that may not normally contract with MnDOT. Methodology for Computing MnDOT FTA DBE Goals FY2016-2018 16 NAICS code within a specific geographic market area. This research product covers some 900 industries, which include about 65 million corporations and other entities (i.e., government agencies, partnerships, non-profits, and educational institutions) and 85 million individuals. For the state of Minnesota, Hoover’s database included information on headquarters, branches and single locations8. Step 2: Queried the Hoover’s database to obtain the list of all the firms, minorityowned firms, and women-owned firms from the relevant NAICS codes in the relevant market areas9. These lists have NAICS codes for each listed firm, along with the firm’s zip code, minority ownership status, and women ownership status. This dataset was used to calculate both the numerator and denominator of the availability rate. Step 3: Narrowly defined the geographic market as PJM-1, PJM-2, and VJM-1. Step 4: Transformed the list into the database where the NAICS code of each firm is the unit of observation, in accordance with the NAICS code-based definition of the prescribed availability formula. Step 5: Calculated the numerator of the D&B availability measure by summing the number of minority-owned business or women-owned enterprises in the geographic market area for each NAICS code. Step 6: Calculated the denominator of the D&B availability measure by summing the number of total firms in in the geographic market area for each NAICS code. Step 7: Calculated the share, or the un-weighted availability, of women-owned business enterprises (WBEs) or minority-owned business enterprises (MBEs), which is the ratio of WBE or MBE firms to total firms in a given relevant industry and narrowly tailored geographic market. Step 8: In some cases the NAICS code for future expenditures produces zero counts of total available firms. The resulting women-owned enterprises or minority-owned business share is not calculable and setting the value to zero provides the erroneous conclusion that there are no available DBEs. Therefore, these cases were treated as missing values, and the values were replaced by the un-weighted mean of the values with positive denominators. “Headquarters” indicates that the company has subsidiaries or branches; “branch” indicates a company location other than the headquarters and single Location indicates that the company does not have any subsidiaries or branches. 9 Unlike the other methods, the D&B method uses “self-reported” minority/gender designations. Thus, the D&B method can include firms that are not MnDOT Certified DBEs. On the other hand, not every certified DBE is included in this database because a requirement of inclusion is the existence of a DUNS number. According to Hoover’s customer service, D&B contacts firms directly to verify their gender or minority status and checks with third party sources and proprietary databases for further verification. 8 Methodology for Computing MnDOT FTA DBE Goals FY2016-2018 17 Step 9: Calculated, for each geographic market area, the weighted availability rate of women-owned enterprises or minority-owned business according to the prescribed availability formula. Step 10: Averaged availability results with and without imputing means in order to obtain the final availability rate for each geographic market area. Please see Tables A-3 in Appendix A for the availability rates for three different GMAs. 3.4.5 The Base Goal The base goal is the weighted average across the two availability methods: Certified DBE List method and Dun & Bradstreet method. Depending on the method used to calculate availability, each defined market area captures a different share of available contract dollars. As a result, each method also yields a different DBE availability goal for each market. In order to derive a single base goal for MnDOT that is based on all the goals calculated for each geographical market, it is necessary to weight each geographical market-specific goal according to the percentage of contract dollars awarded in that area. The two different methods use data that report multiple industries for many of the firms in their databases. We count the firms at the NAICS level. Table 8 presents the details of the calculation of the weighted availability rate at the NAICS code level. This base goal is used in subsequent analysis. Table 8. Computing Base Goal Method PJM-1 PJM-2 VJM1 Unweighted Weighted Average average Certified DBE List 0.83% 1.69% 0.02% Method 5.5367% D&B Method 9.16% 12.21% 9.31% Distribution of Award Amount to compute weights for base goal Percent 95.9% 98.1% 99.2% Distribution of (a) (b) (c) Award Amount 32.7% 33.5% 33.8% Proportional Weight (d) (e) (f) 0.8% Weighted Base Goal 5.5375% 10.2% (d) = (a)/[(a)+(b)+(c)] (e) = (b)/[(a)+(b)+(c)] (f) = (c)/[(a)+(b)+(c)] Note 1: Weighted average is the sum of the availability rates of each GMA multiplied by its proportional weight (d, e, and f), respectively. Note 2: Weighted base goal is the average of weighted averages of all four methods. Methodology for Computing MnDOT FTA DBE Goals FY2016-2018 18 3.5 Adjustment to the Base Goal According to the USDOT regulations, recipients of federal funds are required to adjust their base goals in light of other evidence regarding the market area [49 C.F.R. §26.45(d)]. One valid reason for adjusting the goal would be if the analysis showed discrimination, either in the overall market place or in the specific agency or governmental unit undertaking the procurement and contracting process. The regressions and computations are based on a combined file of FTA and FHWA contract awards from 2012-2014 to DBEs and non-DBEs in the restricted set of NAICs codes. The logic is that MnDOT makes awards in both FTA and FHWA in the same industries. The research team used three methods to measure the degree of discrimination: 1) the dummy interaction variable method, which estimates a coefficient on the DBE dummy variable to control for the relevant observable factors that influence award amount. The purpose of this method is to include the impact FTA funding has on contracts awarded to DBEs. The method enables the research team to capture that impact by adding an interaction term between the variables for DBE and FTA; 2) the Blinder-Oaxaca-Duncan residual difference decomposition estimates separately the contract amounts to DBEs and non-DBEs and computes the amount that DBEs would have received had they been treated like equally situated non-DBEs. The difference between the actual contract amounts and the “equal treatment” amounts defines the discriminatory portion of the gap between DBEs and non-DBEs. In addition to estimating an unexplained portion as a measure for discrimination, an alternative measure of discrimination is computed from a portion of the gap between DBEs and non-DBEs in the FTA variable; and 3) the unadjusted difference in Award Amount between non-DBEs and DBEs due to FTA is calculated by comparing the average award amount between Non-DBEs and DBEs which were weighted by the FTA share of total award amount from the combined FHWA and FTA dataset. 3.5.1 Data Used for Adjustment to the Base Goal The original FTA contract data from MnDOT contains 53 observations, which includes 30 individual vendors that are not identifiable through D&B database, a step that is crucial for acquiring control variables necessary for the regression analysis, thus, the 30 individual vendors were dropped from the data. After dropping the individual vendors, the FTA contract data contains 23 observations with 3 DBEs, which makes it difficult to run any statistical analysis. In order to overcome this limitation, the research team merged the FTA contract data with the FHWA contract data. Then, the team identified the NAICS codes relevant to FTA prime contracts and subcontracts. In the “interaction” of the dataset, there are 693 observations. The summary statistics of the dataset used for the analysis is shown in the summary table. The additional 37 observations (5.3% of the total sample) have dollar values equal to zero for the dependent variable, which is the contract award amount. These observations are dropped from the sample because they are undefined when log-transformed for the final analysis. There were some missing values in a couple of variables, such as credit risk and Methodology for Computing MnDOT FTA DBE Goals FY2016-2018 19 age of firm (missing variable percentages range from 5% to 7%). In order to account for the missing values, extra dummy variables were created for missing observations of the corresponding variables (coded 1 for the missing values and 0 otherwise) and missing values were imputed with a constant value of 1 to retain the full sample. The description of the variables used in the analysis is shown in Table B-1 in Appendix B. The summary statistics of the dataset used in the analysis is shown in Tables B-2 and B-3 in Appendix B for the primes and subcontracts, respectively. The description of the variables used in the analysis is shown in Table B-1 in Appendix B. The summary statistics of the dataset used in the analysis is shown in Table B-2. 3.5.2 Dummy Variable Method The research team developed six statistical models in order to conduct thorough analysis of the FTA and FHWA combined data. The dependent variable is the logarithm of the contract amount. In moving from Model 1 through Model 3, the number of variables being controlled for increases. The variable of interest is the interaction term between DBE and FTA variables (DBE × FTA) where its coefficient indicates incremental discrimination made by FTA in reference to the discrimination made by non-FTA nonDBEs. The models from 1 – 3 use dataset that encompass both prime contracts and subprime contracts whereas models from 4 – 6 use subprime data only. One important thing to notice is that the variable of interest, although the sign is negative, is statistically insignificant throughout different models. These results suggest that FTA-funded project itself, when covariates are controlled, does not contribute to additional discrimination against DBEs. The results are shown in Tables B-3 in Appendix B. 3.5.3 Residual Difference Method Another standard method of measuring discrimination is the residual difference method. The research team used Model 3 of the dummy variable method as a base to run the residual difference method as well as the race neutral portion of DBE goals because they were considered the best models for both groups in terms of goodness of fit. In addition to estimating the portion unexplained as a measure for discrimination, the portion of the gap between DBEs and non-DBEs explained by the FTA is computed as an alternative measure of discrimination. The estimates from these two methods indicates the portion of the gap unexplained after controlling for relevant variables is 60.17%, and the portion of the gap contributed by FTA is 20.13%. The results are detailed in Table B-4 in Appendix B. 3.5.4 Adjustment The base goal is adjusted using the average of the five measures produced from the three above-mentioned methods. The adjustment is 28.79% as shown in Table 9. Methodology for Computing MnDOT FTA DBE Goals FY2016-2018 20 Table 9. Discrimination Gap for Adjustment Method Percent Coefficients on Interaction terms using subs and primes 0.00% Coefficients on Interaction terms using subs 0.00% Oaxaca: Portion of explained gap explained by FTA 20.11% Oaxaca: total unexplained 60.02% UNADJUSTED PERCENT OF GAP DUE TO FTA 63.79% AVERAGE ACROSS METHODS 28.79% * p<0.1, ** p<0.05, *** p<0.01 Add asterisk whenever the difference is statistically significant at 1%, 5% or 10% level, one tail test. 3.6 Race Neutral Analysis In compliance with federal regulations, state and local authorities must identify the maximum feasible portion of the DBE goal that can be achieved through race-neutral measures and the percentage of the goal that can only be achieved through race-conscious measures [49 C.F.R. §26.51(a)]. Myers and Ha have pioneered the use of a detailed econometric procedure that maximizes the race-neutral component of the DBE goals. This method has established a rigorous standard for maximizing the race neutral portion of the overall DBE goal. The logic of the analysis is that some share of DBE dollars awarded would have gone to DBEs without goals. The following method is utilized for the race neutral analysis: To compute the ratio of predicted award dollar amount without goals over predicted award dollar amount with and without goals using the DBE subsample in contract file. Table 10 shows how the race-neutral portion of the goal was calculated. The race neutral analysis based on the above-mentioned method indicates Race-Neutral Percent is 36.07%. Table 10. Race Neutral Analysis Predicted Award Amount without Goals Amount Number of Observations Source: Author's computation Predicted Award Amount with Goals Race Neutral Percent $14,755.07 $26,148.62 36.07% 693 693 Methodology for Computing MnDOT FTA DBE Goals FY2016-2018 21 SECTION 4: THE MnDOT PROPOSED DBE GOALS In summary, the Roy Wilkins Center presented the following recommendation, as shown in Table 11, to the Office of Civil Rights, MnDOT for the proposed overall DBE Goal and race neutral participation. Table 11. Proposed MnDOT FTA DBE Goals FY2016-2018 Base Goal Discrimination Gap for Adjustment Adjusted Goal Race-Neutral Portion Race-Neutral Goal Race-Conscious Portion Race-Conscious Goal 5.54% 28.79% 7.13% 36.07% 2.57% 63.93% 4.56% Source: Author's calculation using the FTA data files. Methodology for Computing MnDOT FTA DBE Goals FY2016-2018 22 SECTION 5: APPENDICES APPENDIX A AVAILABILITY ANALYSIS Table A-1. Weights used for the Availability Analysis NAICS 236220 Description Commercial and Institutional Building Construction Weight 0.00352 237130 Power and Communication Line and Related Structures Construction 0.00022 238110 238120 238130 238140 238150 238160 Poured Concrete Foundation and Structure Contractors Structural Steel and Precast Concrete Contractors Framing Contractors Masonry Contractors Glass and Glazing Contractors Roofing Contractors 0.00643 0.01707 0.00011 0.00086 0.00183 0.00094 238190 Other Foundation, Structure, and Building Exterior Contractors 0.00101 238210 Electrical Contractors and Other Wiring Installation Contractors 0.00698 238220 238310 238320 238350 238390 238910 238990 323111 423130 424710 441310 442110 444130 447110 453210 485111 485113 485310 485410 485510 488410 Plumbing, Heating, and Air-Conditioning Contractors Drywall and Insulation Contractors Painting and Wall Covering Contractors Finish Carpentry Contractors Other Building Finishing Contractors Site Preparation Contractors All Other Specialty Trade Contractors Commercial Printing (except Screen and Books) Tire and Tube Merchant Wholesalers Petroleum Bulk Stations and Terminals Automotive Parts and Accessories Stores Furniture Stores Hardware Stores Gasoline Stations with Convenience Stores Office Supplies and Stationery Stores Mixed Mode Transit Systems Bus and Other Motor Vehicle Transit Systems Taxi Service School and Employee Bus Transportation Charter Bus Industry Motor Vehicle Towing 0.00962 0.00132 0.00043 0.00284 0.00180 0.00396 0.00616 0.01047 0.01196 0.08842 0.03021 0.00008 0.01609 0.08842 0.01331 0.05330 0.09171 0.05437 0.04963 0.02827 0.00460 Methodology for Computing MnDOT FTA DBE Goals FY2016-2018 23 488490 524126 524210 524298 541110 541211 541310 541330 541430 Other Support Activities for Road Transportation Direct Property and Casualty Insurance Carriers Insurance Agencies and Brokerages All Other Insurance Related Activities Offices of Lawyers Offices of Certified Public Accountants Architectural Services Engineering Services Graphic Design Services 0.02684 0.01252 0.01252 0.00050 0.00517 0.00517 0.02419 0.02938 0.01111 541611 Administrative Management and General Mgmt Consulting Services 0.00000 541614 Process, Physical Distribution, and Logistics Consulting Services 0.01319 541690 541850 541990 561320 561499 561720 561730 611519 621112 811111 Other Scientific and Technical Consulting Services Outdoor Advertising All Other Professional, Scientific, and Technical Services Temporary Help Services All Other Business Support Services Janitorial Services Landscaping Services Other Technical and Trade Schools Offices of Physicians, Mental Health Specialists General Automotive Repair 0.00130 0.01111 0.09453 0.05506 0.00273 0.00599 0.00007 0.01272 0.00130 0.04810 811121 Automotive Body, Paint, and Interior Repair and Maintenance 0.01047 812320 Drycleaning and Laundry Services (except CoinOperated) 0.01010 Methodology for Computing MnDOT FTA DBE Goals FY2016-2018 24 PMJ-1 NAICS Weight 236220 237130 238110 238120 238130 238140 238150 238160 238190 238210 238220 238310 238320 238350 238390 238910 238990 423130 424710 441310 442110 444130 447110 453210 485113 488410 524126 524210 524298 541110 541211 541430 541611 541690 541850 541990 0.00579 0.00037 0.01058 0.02806 0.00017 0.00141 0.00301 0.00154 0.00167 0.01147 0.01582 0.00217 0.00071 0.00467 0.00297 0.00651 0.01013 0.01966 0.14536 0.04967 0.00013 0.02645 0.14536 0.02188 0.15076 0.00756 0.02059 0.02059 0.00082 0.00849 0.00849 0.01826 0.00000 0.00214 0.01826 0.09983 Number of DBE firms 15 4 12 6 7 11 5 8 9 25 13 15 22 23 8 39 25 2 1 1 3 1 11 27 11 1 6 Total Number of Firms Unweighted Rate Weighted Rate 704 141 540 50 332 509 78 333 96 1570 1725 488 694 592 193 894 1220 42 74 729 378 476 1888 103 23 117 284 3254 61 2428 969 463 1307 366 71 476 0.0213 0.0284 0.0222 0.1200 0.0211 0.0216 0.0641 0.0240 0.0938 0.0159 0.0075 0.0307 0.0317 0.0389 0.0415 0.0436 0.0205 0.0000 0.0000 0.0000 0.0053 0.0021 0.0000 0.0097 0.0000 0.0000 0.0000 0.0000 0.0000 0.0012 0.0010 0.0238 0.0207 0.0301 0.0141 0.0126 0.0001 0.0000 0.0002 0.0034 0.0000 0.0000 0.0002 0.0000 0.0002 0.0002 0.0001 0.0001 0.0000 0.0002 0.0001 0.0003 0.0002 0.0000 0.0000 0.0000 0.0000 0.0001 0.0000 0.0002 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0004 0.0000 0.0001 0.0003 0.0013 Methodology for Computing MnDOT FTA DBE Goals FY2016-2018 25 561720 561730 611519 621112 811111 812320 Total 0.00984 0.00011 0.02091 0.00214 0.07908 0.01661 1.00000 15 31 1 1 359 1084 1955 47 125 1647 212 28738 0.0138 0.0159 0.0213 0.0000 0.0006 0.0000 0.0125 Methodology for Computing MnDOT FTA DBE Goals FY2016-2018 0.0001 0.0000 0.0004 0.0000 0.0000 0.0000 0.0083 26 PMJ-2 NAICS Weight 236220 237130 238110 238120 238130 238140 238150 238160 238190 238210 238220 238310 238320 238350 238390 238910 238990 423130 424710 441310 442110 444130 447110 453210 485113 488410 524126 524210 524298 541110 541211 541430 541611 541690 541850 541990 0.00579 0.00037 0.01058 0.02806 0.00017 0.00141 0.00301 0.00154 0.00167 0.01147 0.01582 0.00217 0.00071 0.00467 0.00297 0.00651 0.01013 0.01966 0.14536 0.04967 0.00013 0.02645 0.14536 0.02188 0.15076 0.00756 0.02059 0.02059 0.00082 0.00849 0.00849 0.01826 0.00000 0.00214 0.01826 0.09983 Number of DBE firms 8 1 3 3 6 8 4 8 5 20 10 7 14 14 5 13 15 1 3 1 9 21 10 1 7 Total Number of Firms Unweighted Rate Weighted Rate 264 29 125 10 77 146 35 129 20 473 526 135 310 190 78 217 359 19 17 223 167 146 632 48 8 43 99 1317 36 1518 478 319 820 191 43 273 0.0303 0.0345 0.0240 0.3000 0.0779 0.0548 0.1143 0.0620 0.2500 0.0423 0.0190 0.0519 0.0452 0.0737 0.0641 0.0599 0.0418 0.0000 0.0000 0.0000 0.0060 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0020 0.0021 0.0282 0.0256 0.0524 0.0233 0.0256 0.0002 0.0000 0.0003 0.0084 0.0000 0.0001 0.0003 0.0001 0.0004 0.0005 0.0003 0.0001 0.0000 0.0003 0.0002 0.0004 0.0004 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0005 0.0000 0.0001 0.0004 0.0026 Methodology for Computing MnDOT FTA DBE Goals FY2016-2018 27 561720 561730 611519 621112 811111 812320 Total 0.00984 0.00011 0.02091 0.00214 0.07908 0.01661 1.00000 15 18 1 1 232 516 665 27 92 605 119 11544 0.0291 0.0271 0.0370 0.0000 0.0017 0.0000 0.0201 Methodology for Computing MnDOT FTA DBE Goals FY2016-2018 0.0003 0.0000 0.0008 0.0000 0.0001 0.0000 0.0169 28 VJM-1 NAICS Weight 236220 237130 238110 238120 238130 238140 238150 238160 238190 238210 238220 238310 238320 238350 238390 238910 238990 423130 424710 441310 442110 444130 447110 453210 485113 488410 524126 524210 524298 541110 541211 541430 541611 541690 541850 541990 0.00579 0.00037 0.01058 0.02806 0.00017 0.00141 0.00301 0.00154 0.00167 0.01147 0.01582 0.00217 0.00071 0.00467 0.00297 0.00651 0.01013 0.01966 0.14536 0.04967 0.00013 0.02645 0.14536 0.02188 0.15076 0.00756 0.02059 0.02059 0.00082 0.00849 0.00849 0.01826 0.00000 0.00214 0.01826 0.09983 Number of DBE firms 1 1 Total Number of Firms Unweighted Rate Weighted Rate 45 10 27 1 9 24 7 11 4 70 84 20 33 17 8 29 35 1 5 50 23 27 88 9 1 5 27 254 6 414 156 41 202 64 7 41 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0119 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0050 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0002 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 Methodology for Computing MnDOT FTA DBE Goals FY2016-2018 29 561720 561730 611519 621112 811111 812320 Total 0.00984 0.00011 0.02091 0.00214 0.07908 0.01661 1.00000 2 55 60 5 22 69 22 2088 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0010 Methodology for Computing MnDOT FTA DBE Goals FY2016-2018 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0002 30 Table A-3. Availability Analysis - D&B Method, All NAICS PMJ-1 NAICS Weight Number of DBE firms Total Number of Firms Unweighted Rate Weighted Rate 236220 237130 238110 238120 238130 238140 238150 238160 238190 238210 238220 238310 238320 238350 238390 238910 238990 423130 424710 441310 442110 444130 447110 453210 485113 488410 524126 524210 524298 541110 541211 541430 541611 541690 541850 0.0058 0.0004 0.0106 0.0281 0.0002 0.0014 0.0030 0.0015 0.0017 0.0115 0.0158 0.0022 0.0007 0.0047 0.0030 0.0065 0.0101 0.0197 0.1454 0.0497 0.0001 0.0264 0.1454 0.0219 0.1508 0.0076 0.0206 0.0206 0.0008 0.0085 0.0085 0.0183 0.0000 0.0021 0.0183 344 13 78 21 54 65 22 71 5 280 261 59 224 18 11 115 239 10 4 40 124 75 0 53 5 24 10 446 18 685 126 525 1,505 404 13 2,272 87 1,173 67 1,338 928 189 1,031 29 3,241 4,572 1,351 2,930 360 255 1,560 3,267 120 199 866 973 929 6 224 29 491 166 5,411 192 9,014 1,376 1,725 7,270 1,669 95 0.1514 0.1494 0.0665 0.3134 0.0404 0.0700 0.1164 0.0689 0.1724 0.0864 0.0571 0.0437 0.0765 0.0500 0.0431 0.0737 0.0732 0.0833 0.0201 0.0462 0.1274 0.0807 0.0000 0.2366 0.1724 0.0489 0.0602 0.0824 0.0938 0.0760 0.0916 0.3043 0.2070 0.2421 0.1368 0.0009 0.0001 0.0007 0.0088 0.0000 0.0001 0.0004 0.0001 0.0003 0.0010 0.0009 0.0001 0.0001 0.0002 0.0001 0.0005 0.0007 0.0016 0.0029 0.0023 0.0000 0.0021 0.0000 0.0052 0.0260 0.0004 0.0012 0.0017 0.0001 0.0006 0.0008 0.0056 0.0000 0.0005 0.0025 Methodology for Computing MnDOT FTA DBE Goals FY2016-2018 31 541990 561720 561730 611519 621112 811111 812320 Total 0.0998 0.0098 0.0001 0.0209 0.0021 0.0791 0.0166 1.0000 870 388 181 50 44 161 170 7,811 10,625 2,089 3,354 215 435 3,307 676 76,106 0.0819 0.1857 0.0540 0.2326 0.1011 0.0487 0.2515 0.0082 0.0018 0.0000 0.0049 0.0002 0.0038 0.0042 0.1026 0.0916 Methodology for Computing MnDOT FTA DBE Goals FY2016-2018 32 PMJ-2 NAICS Weight Number of DBE firms Total Number of Firms Unweighted Rate Weighted Rate 236220 237130 238110 238120 238130 238140 238150 238160 238190 238210 238220 238310 238320 238350 238390 238910 238990 423130 424710 441310 442110 444130 447110 453210 485113 488410 524126 524210 524298 541110 541211 541430 541611 541690 541850 541990 0.0058 0.0004 0.0106 0.0281 0.0002 0.0014 0.0030 0.0015 0.0017 0.0115 0.0158 0.0022 0.0007 0.0047 0.0030 0.0065 0.0101 0.0197 0.1454 0.0497 0.0001 0.0264 0.1454 0.0219 0.1508 0.0076 0.0206 0.0206 0.0008 0.0085 0.0085 0.0183 0.0000 0.0021 0.0183 0.0998 258 10 47 5 28 41 13 47 1 182 160 34 155 12 9 34 145 6 2 20 77 42 0 41 4 13 5 237 11 595 90 417 1,334 358 10 678 1,272 37 376 21 543 335 103 480 11 1,419 1,967 434 1,507 127 113 380 1,527 63 30 293 536 328 2 144 15 259 102 2,674 131 7,197 1,034 1,274 5,650 1,255 72 6,417 0.2028 0.2703 0.1250 0.2381 0.0516 0.1224 0.1262 0.0979 0.0909 0.1283 0.0813 0.0783 0.1029 0.0945 0.0796 0.0895 0.0950 0.0952 0.0667 0.0683 0.1437 0.1280 0.0000 0.2847 0.2667 0.0502 0.0490 0.0886 0.0840 0.0827 0.0870 0.3273 0.2361 0.2853 0.1389 0.1057 0.0012 0.0001 0.0013 0.0067 0.0000 0.0002 0.0004 0.0002 0.0002 0.0015 0.0013 0.0002 0.0001 0.0004 0.0002 0.0006 0.0010 0.0019 0.0097 0.0034 0.0000 0.0034 0.0000 0.0062 0.0402 0.0004 0.0010 0.0018 0.0001 0.0007 0.0007 0.0060 0.0000 0.0006 0.0025 0.0105 Methodology for Computing MnDOT FTA DBE Goals FY2016-2018 33 561720 561730 611519 621112 811111 812320 Total 0.0098 0.0001 0.0209 0.0021 0.0791 0.0166 1.0000 271 97 38 40 108 151 5,826 1,159 1,282 167 339 1,544 530 43,149 0.2338 0.0757 0.2275 0.1180 0.0699 0.2849 0.0023 0.0000 0.0048 0.0003 0.0055 0.0047 0.1350 0.1221 Methodology for Computing MnDOT FTA DBE Goals FY2016-2018 34 VJM-1 NAICS Weight Number of DBE firms Total Number of Firms 236220 237130 238110 238120 238130 238140 238150 238160 238190 238210 238220 238310 238320 238350 238390 238910 238990 423130 424710 441310 442110 444130 447110 453210 485113 488410 524126 524210 524298 541110 541211 541430 541611 541690 541850 541990 0.0058 0.0004 0.0106 0.0281 0.0002 0.0014 0.0030 0.0015 0.0017 0.0115 0.0158 0.0022 0.0007 0.0047 0.0030 0.0065 0.0101 0.0197 0.1454 0.0497 0.0001 0.0264 0.1454 0.0219 0.1508 0.0076 0.0206 0.0206 0.0008 0.0085 0.0085 0.0183 0.0000 0.0021 0.0183 0.0998 10 0 4 0 3 5 2 2 0 8 8 3 5 1 0 4 10 0 0 3 10 2 0 2 1 0 1 20 1 63 10 31 97 29 1 54 130 2 39 1 58 39 25 47 0 123 180 44 90 10 16 48 137 7 9 45 51 42 0 16 3 15 10 341 12 800 133 109 487 130 7 463 Unweighted Rate Weighted Rate 0.0769 0.0000 0.1026 0.0000 0.0517 0.1282 0.0800 0.0426 0.0004 0.0000 0.0011 0.0000 0.0000 0.0002 0.0002 0.0001 0.0650 0.0444 0.0682 0.0556 0.1000 0.0000 0.0833 0.0730 0.0000 0.0000 0.0667 0.1961 0.0476 0.0007 0.0007 0.0001 0.0000 0.0005 0.0000 0.0005 0.0007 0.0000 0.0000 0.0033 0.0000 0.0013 0.1250 0.3333 0.0000 0.1000 0.0587 0.0833 0.0788 0.0752 0.2844 0.1992 0.2231 0.1429 0.1166 0.0027 0.0503 0.0000 0.0021 0.0012 0.0001 0.0007 0.0006 0.0052 0.0000 0.0005 0.0026 0.0116 Methodology for Computing MnDOT FTA DBE Goals FY2016-2018 35 561720 561730 611519 621112 811111 812320 Total 0.0098 0.0001 0.0209 0.0021 0.0791 0.0166 1.0000 14 5 0 3 0 7 419 68 123 11 22 131 36 4,060 0.2059 0.0407 0.0000 0.1364 0.0000 0.1944 0.0020 0.0000 0.0000 0.0003 0.0000 0.0032 0.1032 0.0931 Methodology for Computing MnDOT FTA DBE Goals FY2016-2018 36 APPENDIX B DISCRIMINATION ANALYSIS Table B-1. Description of Variables in the Regression Analysis Variable Description DBE Dummy variable to indicate DBE status FTA Dummy variable to indicate FTA status FFY2012 Dummy variable for federal fiscal year 2012 FFY2013 Dummy variable for federal fiscal year 2013 FFY2014 Dummy variable for federal fiscal year 2014 Credit2 Dummy variable for low credit (from Dun&Brad) Credit3 Dummy variable for medium credit (from Dun&Brad) Credit4 Dummy variable for high credit (from Dun&Brad) Missing_credit_risk Dummy for missing credit risk Minnesota Dummy variable for location in Minnesota Large firm Dummy variable for firms with more than 100 employees and total sales greater than $1 million Firm_age Continuous variable for age of firm Missing_firm_age Dummy for missing firm age ArchiEnginServ Dummy: NAICS defined industry for architectural engineering service ManaScienTechnicServ Dummy: NAICS defined industry for foundation for management science technical service Source: MnDOT FTA contracts: Oct. 1, 2011 - Sep. 30, 2014 Methodology for Computing MnDOT FTA DBE Goals FY2016-2018 37 Table B-2. Summary Statistics for Subcontracts Variable Mean Std. Dev. Min. Max. ln award amount 10.182 1.497 5.293 15.193 DBE 0.229 0.421 0 1 FTA 0.033 0.179 0 1 DBE FTA interaction 0.004 0.066 0 1 FFY2013 0.388 0.488 0 1 FFY2014 0.231 0.422 0 1 minnesota 0.97 0.172 0 1 ArchiEnginServ 0.199 0.4 0 1 ManaScienTechnicServ 0.039 0.194 0 1 credit risk3 0.039 0.194 0 1 Credit Risk 4 0.043 0.204 0 1 missing_credit_risk 0.068 0.252 0 1 large firm2 0.052 0.222 0 1 firm age 20.413 16.436 0 100 missing firm age 0.049 0.216 0 1 N 693 Source: MnDOT FTA contracts: Oct. 1, 2011 - Sep. 30, 2014 Methodology for Computing MnDOT FTA DBE Goals FY2016-2018 38 Table B-3. Discrimination Analysis Dummy Interaction Variable Method for Prime Contracts: By Natural Log (ln) Award Amount Both Primes and Sub data DBE_FTA_interaction Prime FFY2013 FFY2014 minnesota ArchiEnginServ ManaScienTechnicServ credit_risk3 credit_risk4 missing_credit_risk (1) (2) (3) (4) (5) (6) ln_AA ln_AA ln_AA ln_AA ln_AA ln_AA -1.372 -1.294 -0.996 -1.452 -1.388 -1.452 (1.421) (1.434) (1.466) (1.381) (1.394) (1.381) 2.440*** 2.419*** 2.668*** (0.205) (0.206) (0.265) 0.237* 0.249* 0.234* 0.236* 0.246* 0.236* (0.127) (0.13) (0.128) (0.133) (0.135) (0.133) 0.151 0.154 0.151 0.125 0.126 0.125 (0.138) (0.138) (0.138) (0.142) (0.142) (0.142) 0.14 0.177 0.119 -0.043 -0.00967 -0.043 (0.465) (0.461) (0.461) (0.483) (0.479) (0.483) 0.155 0.161 0.151 0.183 0.189 0.183 (0.134) (0.134) (0.134) (0.143) (0.143) (0.143) 0.28 0.309 0.277 0.212 0.238 0.212 (0.608) (0.601) (0.608) (0.634) (0.627) (0.634) 0.0291 0.00893 0.0285 0.0418 0.0255 0.0418 (0.279) (0.28) (0.279) (0.28) (0.281) (0.28) 0.672* 0.661* 0.633* 0.829* 0.818* 0.829* (0.369) (0.368) (0.38) (0.43) (0.43) (0.43) 0.15 0.17 0.159 0.0714 0.0886 0.0714 (0.255) (0.255) (0.257) (0.263) (0.264) (0.263) DBE -0.0919 -0.0741 (0.129) (0.129) FTA _cons Subs Only data -0.392 0 (0.377) (.) 9.722*** 9.701*** 9.745*** 9.903*** 9.882*** 9.903*** -0.465 -0.462 -0.462 -0.482 -0.479 -0.482 656 656 656 621 621 621 adj. R2 0.153 0.153 0.153 0.01 0.008 0.01 F 18.16 16.66 19.41 1.077 1.011 1.077 N Standard errors in parentheses * p<0.1, ** p<0.05, *** p<0.01 Methodology for Computing MnDOT FTA DBE Goals FY2016-2018 39 Table B-4. Residual Difference Analysis using Oaxaca Decomposition (1) ln_award_amount Differential Prediction_1 Prediction_2 Difference Explained FTA FHWAprime FFY2013 FFY2014 minnesota ArchiEnginServ ManaScienTechnicServ credit_risk3 credit_risk4 missing_credit_risk large_firm2 firm_age missing_firm_age Total 10.23*** (0.068) 10.01*** (0.121) 0.226 (0.139) 0.0455 (0.033) 0.0600** (0.0286) -0.0152 (0.0242) 0.00742 (0.0112) 0.00062 (0.00777) 0.012 (0.0127) 0.000893 (0.0033) 0.00407 (0.0119) 0.0164 (0.0138) -0.00232 (0.0292) 0.00235 (0.0168) 0.000377 (0.00202) -0.0417 (0.0299) 0.0905 (0.0681) Unexplained Methodology for Computing MnDOT FTA DBE Goals FY2016-2018 40 FTA 0.110* (0.0638) 0.00706 (0.00743) -0.276* (0.164) -0.0457 (0.0639) 2.928** (1.437) -0.235*** (0.0897) -0.0194 (0.0496) 0.00994 (0.0102) -0.00262 (0.0135) 0.0223 (0.089) 0.0248 (0.0209) 0.239 (0.22) 0.116 (0.0752) -2.743* (1.487) 0.136 (0.152) 656 FHWAprime FFY2013 FFY2014 minnesota ArchiEnginServ ManaScienTechnicServ credit_risk3 credit_risk4 missing_credit_risk large_firm2 firm_age missing_firm_age _cons Total N adj. R2 F Standard errors in parentheses * p<0.1, ** p<0.05, *** p<0.01 Unexplained Portion is 60.17% Explained by FTA is 20.13% Methodology for Computing MnDOT FTA DBE Goals FY2016-2018 41