A featured contribution from Leadership Perspectives: a curated forum reserved for leaders nominated by our subscribers and vetted by the Construction Tech Review Advisory Board.

QED Investors

Amias Gerety, Partner

Construction Risk is a Data Problem

Amias Gerety

Amias Gerety

Construction lending is a high-risk business. It is defined by lending large sums where the collateral for the loan is yet to be built. To play with the line from the Kevin Costner classic Field of Dreams—it’s not “if you build it they will come” but “if we build it, you’ll be secured.” 

All around us, the risks for this type of lending have never been more evident. Though construction is considered an “essential activity” in most states, even during COVID-induced shutdowns, construction lending actually increased in Q2 of this year. Nobody knows what the future of cities will look like after this pandemic, and nobody knows when or how the wave of real estate-related bankruptcies will ripple through the market. 

Moreover, construction—even construction lending at the largest banks in the world – is an analogue business. Risk management practices for lenders, developers and equity partners all rely on human beings to review hundreds or thousands of pages of PDF images or printed pages, compare those pages to budgets and plans and hope that they didn’t miss something significant. Though there are processes to control and review these plans, these controls have very little data and almost no analysis to support them. Instead, they are underpinned by trust – trust in business partners and trust in employees who “stare and compare” and try to understand what’s happening in a construction project.

This will change. Better risk management, better monitoring, and most importantly, better business decisions are possible if we modernize the data infrastructure for construction lending.

In the world of data science, big opportunities are usually defined by simple criteria: most importantly, there have to be business decisions that are frequently and materially incorrect; and then there has to be a relevant data set that can be analyzed to improve those business decisions.

In the financial crisis, loss rates on construction lending peaked at 8 percent, so we definitely have a rich pool of business decisions that proved to be incorrect in hindsight. And now with advances in computer vision and document analysis, the underlying data in construction processes can be captured, centralized and analyzed. 

“2020 has underscored for all of us that the practice of risk management is defined by its unpredictability.”

For example, our portfolio company, Rabbet can automatically parse a full draw request – sorting invoices and lien releases from inspection reports and identifying how those documents relate to individual line items in the budget. Their software can automatically identify potential errors, underspend/overspend, and benchmark data against other construction projects in your portfolio or the industry more generally.

Once that data is captured, the analysis is both shocking and straightforward. For example, in early March, Rabbet found that 4 out of 5 projects that they were onboarding to their system had material issues, and one draw request alone had a million-dollar mistake. And many of these errors were found by simply asking the computer to compare data taken from invoices to data entered by the developer in the excel spreadsheet summary.

All risk management must include identification of risks, decisions about whether to accept the risks and steps to mitigate. All of these steps are not possible without an automated approach to centralize and analyze the data about construction projects. 

The articles from these contributors are based on their personal expertise and viewpoints, and do not necessarily reflect the opinions of their employers or affiliated organizations.
Top