From the kitchen table to the reporting schedule

Mandate

Make the bank's filed, tested and modelled figures traceable to accountable records.

Outcomes

• Named owners and written definitions for 1,140 critical data elements
• Amended two call reports and added a quarterly collateral reconciliation
• Built pricing-exception reason codes and monthly exception reporting
• Negotiated data rights into the 2024 core renewal

Decisions in context

I took this role when the bank decided that data ownership needed one accountable executive below the chief operating officer. I lead four teams: governance and quality, regulatory-reporting data, analytics and model data, and fair-lending analytics. My budget covers people, tools and external validation work; core processing and infrastructure belong to technology. The first year went on agreeing owners for 1,140 critical data elements, each with a named business owner and a written definition. Since then the work has been to keep those definitions alive as products, systems and rules change. The three cases on this site sit within this period. I present the controls behind the call report to the audit committee, and fair-lending results to the risk committee with the chief compliance officer. I am an officer of the bank, not a director, and I do not vote on credit.

Reason for the move

Current appointment; exploring one compatible external board seat.

Mandate

Establish data ownership and an internal, reproducible basis for fair-lending testing.

Outcomes

• Set up the bank's data governance council
• Brought annual fair-lending testing in-house
• Documented testing methods so examiners could rerun them

Decisions in context

Before 2015, fair-lending testing at Kellanbrook relied on an outside firm's annual review of a sample. Theron Vasbinder, newly arrived as chief compliance officer, wanted the bank to understand its own patterns between reviews. I built a small team to do that and, in parallel, set up the data governance council. The two jobs informed each other. Testing pricing on agricultural lines exposed how inconsistently lenders recorded the reasons for an exception, and that inconsistency became one of the first quality issues the council took on. Above all, I learnt to separate the analysis from the conclusion: my team produced reproducible tests, documented for examiners, and compliance decided what they meant. We also produced our first proxy-based estimates, and I saw how quickly their reliability falls away in small rural branches with few applicants.

Reason for the move

Appointed chief data officer when the bank consolidated data ownership under one executive.

Mandate

Give management and finance one reconciled set of loan and deposit figures.

Outcomes

• Rebuilt the management pack from the call report's reconciled data
• Documented every deliberate difference between internal and regulatory segments
• Reduced quarter-end reconciling items from about 300 to under 40

Decisions in context

Kellanbrook hired me after a period in which management reports and the call report described the loan book differently and nobody could fully explain why. I reported to the chief financial officer and led six analysts. We rebuilt the monthly management pack from the same reconciled data that fed the call report, and we documented each place where the two deliberately differed, such as internal segments that cut across regulatory categories. I also learnt the bank's lending culture, sitting with agricultural lenders in branch offices to see how a request became a loan record. That time mattered later. When my team found a classification problem in 2021, I knew the people who had made the entries and why the core had allowed them, which kept the conversation about records rather than about individual blame.

Reason for the move

Asked to lead data governance and build a fair-lending analytics capability.

Mandate

Move each converting bank's records onto the platform with their meaning intact.

Outcomes

• Led the data work on 41 client conversions
• Introduced a pre-conversion report of ambiguous codes for each client
• Wrote the team's guide to mapping collateral and purpose codes

Decisions in context

At Arbelune I led the data side of core conversions, taking a bank's loans, deposits, collateral and customers from its old system onto ours, usually across one weekend. I worked on 41 conversions, the later ones as manager of a nine-person team. Each began with a mapping exercise that revealed how much meaning lived outside the fields: codes repurposed years earlier, notes holding the real collateral description, and purpose codes used differently between branches. My job was to make those differences visible to the client before the weekend, not to resolve them by guesswork. I also saw the vendor's side of release management, and how a definition change that looked minor to developers could alter a client's regulatory figures. I left because I wanted to work where data decisions had consequences I would own myself, inside one bank.

Reason for the move

Wanted to own the consequences of data decisions inside a bank rather than hand them back to clients.

Mandate

Prepare the loan and deposit schedules of the call report and explain their movements.

Outcomes

• Prepared the loan and deposit schedules for 25 quarterly filings
• Started a classification log later adopted as the reference for new analysts
• Traced quarter-on-quarter movements to individual loans

Decisions in context

My work at Corvenhall was the quarterly call report, mainly its loan and deposit schedules. The bank financed feedlots, grain operations and Nebraska businesses, and its loan categories depended on choices made in branches I never visited. I spent much of each quarter tracing movements back to individual loans and asking lenders why a code had been used. Most answers were reasonable; some were simply habit. I began keeping a log of classification questions and their resolutions, which the controller later adopted as the reference for new analysts. During these years I studied part-time for a master's degree in applied statistics, completed in 2003, which gave me the methods I later used in fairness testing. From 2003 I supervised two analysts. I left to see those systems from the inside, at the vendor that built them.

Reason for the move

Joined a core vendor to understand where the records I reported actually originated.

Mandate

Reconcile member records and prepare comparable annual benchmarks.

Outcomes

• Kept records for about seventy wheat, sorghum and cow-calf operations
• Prepared annual cost-per-acre and cost-per-head benchmarks
• Flagged records that could not support a benchmark comparison

Decisions in context

I joined the association straight from college. Members paid for help keeping their farm records and for annual benchmarks comparing their costs and returns with similar operations. I was assigned about seventy farms, mostly wheat, sorghum and cow-calf operations, and visited each at least twice a year. The work taught me that a correct record can still mislead. A land payment booked as an operating expense, a crop held over the year-end, or a machinery trade valued at the dealer's figure could each distort a benchmark without any number being wrong. I learnt to ask how an entry was made before deciding what it meant, and to tell members when their records could not support the comparison they wanted. I left for bank reporting because I wanted the same questions at a scale where many people relied on the answer.

Reason for the move

Moved into bank regulatory reporting to work on records that many more people relied upon.