New research about mortgage brokersTom LaMalfaWhat is a mortgage broker? Wholesale Access recently published "Mortgage Brokers 2004," our seventh major examination of the mortgage brokerage industry since 1991. This article reviews the broker studies, their focus, what we learned, their methodologies, statistical validity, consistency, measurement, survey design and the objectives of the research. The series began with "The Nation's Mortgage Brokers—Who are they? Where are they? What do they do?" Study number two was an examination of the economics of brokering mortgages, "The Cost of Origination—Are mortgage brokers the low cost producers?" That research was completed in 1992. The third study was completed in 1995. It looked at how mortgage brokerages were adapting to a major downturn in origination activity. There had been a tremendous wave of refinance activity in late 1992 through 1993 that swelled production. We wanted to know how brokerages were fairing, coping with a near 40 percent drop in originations between late 1993 and mid-year 1994. Studies four to seven were cleverly titled "Mortgage Brokers 1998," "Mortgage Brokers 2000," "Mortgage Brokers 2002" and now (big surprise), the latest—"Mortgage Brokers 2004." The serials all focused on five areas of the brokerages' business and industry: technology, wholesalers, operations, products and demographics. The technology section examined in some detail the software used, automated underwriting system (AUS) usage and use of the Internet. The section on wholesalers looks at who brokerages sell to, in what quantity/ranking and why. The operations section deals with sourcing business, locking, loan officer splits, overages, document preparation and closing. The section on products looks at five loan types: agency (conforming), jumbo, alt-A, FHA/VA and sub-prime. Here, our interest as researchers was longitudinal—how is the product market changing over time? The final section addresses demographics: where brokerages are located, production volume, number of employees, brokers' educational backgrounds, association affiliations, what publications brokers read, etc. So, what did we learn? The complete answer takes 380 pages of statistical tables, analyses and addendums. However, key highlights would include the following. First, we estimate that 53,000 brokerage firms were operating in 2004. (Wholesale Access uses the term broker to mean the firm's owner[s]. Some brokerages have three or even four partners or co-owners.) In 2002, we counted 44,000 firms, 17 percent fewer than two years earlier. Second, brokerages are getting smaller. Since 1998, the mean firm size slipped from 9.3 to 7.9 employees, while the median has declined from five to four employees. Back in 1991, the median was six employees. The industry's fragmentation and the smaller firm size are particularly interesting because they contrast sharply from most other industries where consolidation has occurred—from banks to pharmaceuticals to defense contractors. In most industries, firm size has risen—fewer, but bigger banks for example. The third important thing learned was that the brokerage industry's market share was 68 percent in 2004. This was up three percentage points from 2002's estimate. Brokerages' share of residential originations has exceeded 60 percent every year since 2001. Their market share has been on the ascent since the late 1980s. In 2000, we had been expecting a pullback in market share to about 50 percent for the brokerage industry in 2005. This obviously hasn't occurred. The explanation for our failed forecast is that refinance activity has held at historically high levels—about 50 percent of aggregate originations—and the product menu has shifted, favoring the nimbler players. So a key question is: Will the industry remain as competitive to account for two-thirds rather than one-half of origination volume in 2006? More competition is coming from real estate agents and homebuilders—the traditional sources of business for brokers—who have been increasingly setting up mortgage finance operations. Additionally, the largest wholesalers are expanding their retail franchises significantly. It appears that many more resources have been put into retail channels in recent years by some of the largest wholesalers. That suggests more competition coming down the road for brokerages from wholesale lenders. Worth discussing at this point is the brokerage study's methodology and, importantly, its validity. All research is not created equal. "Mortgage Brokers 2004" was a survey research project. The only way to collect the data that we wanted was to ask brokers what we wanted to know. We have to get them to want to talk with our callers. To a group of 1,000 brokers, we asked 73 questions with multiple parts. To a second group of 1,500, we asked 16 questions. We made almost 70,000 calls to verify firm names, addresses and phone numbers, thereby verifying and inventorying approximately 45,000 (or 85 percent of our estimated 53,000) operating brokerages. All surveys were conducted between February and June of 2005. Except for the firms tested in the pilot phase, all surveys were conducted by telephone of brokers chosen randomly from a list of 53,000 firms. Respondents were assured of confidentiality—another requirement in this type of survey research. It was dialing for data from Wholesale Access' sampling frame, an inventory of firm names, addresses and phone numbers we've accumulated over the years. Wholesale Access frequently inputs new names to add and purges the list with the goal of keeping it up-to-date between studies. The number of surveys completed per state was relatively proportional to the size of the state market. Randomness is a critical factor in creating statistical validity. Equally important is sample size. A random number of about 350 responses to a single question get to the minimum validity threshold. Having more surveys increases the confidence level. As stated, we long-surveyed about 1,000 brokerages. Statistical validity is necessary in order to have data that can (if needed) be used in regulatory, legislative and judicial settings to explain the business and industry to laymen, whether media, congressional or other. Indeed, only statistically valid studies are of any value from a decision-making perspective. Non-random surveys and surveys that are not well defined are near useless or should be for decision makers. For example, the recent Campbell/Inside Mortgage Finance (IMF) broker survey is voluminous but not statistically valid, since anyone could answer it—whether a broker or not—and there could be multiple respondents from the same firm. Most of their respondents were individual loan officers, whereas our respondents are the owners of the firms and their answers reflect the volume for that firm. All of the answers are then weighed by the volume of each firm. Wholesale Access is highly focused on the ideastream that emanates from collected and compiled data. Most importantly, are the findings consistent over time and from study to study, as well as with other research? What's the story? How volatile is that main story? Does it make sense when juxtaposed and examined against the observed, non-empirical market environment? Having mentioned IMF, it is important to note that the mortgage publication measures broker business differently than Wholesale Access. We ask who took the application from the borrower and processed the loan. IMF (and National Mortgage News) tallies from lenders the amount of business coming through each of three production channels. This method of measuring overlooks the originations of brokerages that land in the correspondent and retail channel columns. The latter two channels aren't pure by any stretch. Brokerages originate loans that get counted in each of these channels. Our research over the years indicates that about 50 percent of the business counted in the correspondent channel and 20 percent of the business counted in the retail channel was actually originated by brokerages. They were at the point of sale taking the borrower's (or refinancer's) application and processing the loan while determining to which wholesaler the loan should be sold. I'll conclude with some thoughts on survey design and our collective objectives from the research. First, according to Peter F. Drucker, America's management guru and the so-called "Sage of Claremont," the two most important types of research for firms in an industry are customer research and market research. Customer research is assembling a database about the customer, his likes/dislikes, etc. Market research involves developing a database about elements of the marketplace—of products, lenders or technology, for example. "Mortgage Brokers 2004," like its predecessors, was designed to provide databases about customers and markets. Moreover, the studies also include an evaluative analysis component. Major wholesalers can measure how they are viewed by brokers in an array of areas and how brokers view their major competitors. For instance, Countrywide can ascertain what brokers report about them as well as about Wells Fargo or InterFirst, etc. So, customer data, market data and competitive analysis are baked into the cake. Our second goal was to provide data and useful analysis to a half dozen functional areas within the wholesaler's firm. As a result, data is collected for production, for the information technology department, for operations, for strategic planning, for marketing, for legal and, of course, for senior management. For example, the technology and Internet section of the survey instrument asked: How does the brokerage use the Internet to communicate with borrowers and lenders? How do they decide what loans get submitted to which AUS? What loan origination system and customer relationship management software they use? Information technology personnel don't have to make systems integration decisions based on anecdotal information, quick and dirty surveys of five or six of their firm's brokerage customers or guesstimates. Hard data is available. The report also measured the volume of loans by all of the major mortgage wholesalers in the industry, as well as the perceptions of service given by these wholesalers to their brokers. All of this data was consistent with our studies of lenders and the HMDA database. We make sure that all of our research is mutually consistent to double check its validity. I would like to thank the firms that gave us the funds to do the study: American Home Mortgage, American Mortgage Network, Bank of America Mortgage, Bernstein-Rein Group, Calyx Software, Chase Home Finance, Citicorp Mortgage, Countrywide Home Loans, Ellie Mae, Fannie Mae, First Magnus, GMAC/RFC, HSBC/Decision One, IndyMac Bank, InterFirst/ABN Amro, Loan City Mortgage, MGIC, National City Mortgage, Scotsman Guide, SunTrust Mortgage and Wells Fargo Home Mortgage. We can't bring good, solid data to the industry without your continued support. Thank you. Tom LaMalfa is co-founder, partner and managing director of Wholesale Access Mortgage Research & Consulting Inc., which focuses on mortgage finance and does market structure studies, production revenue, expense benchmarking, multi-client projects and proprietary research. He can be reached at (216) 752-6610 or e-mail [email protected].