NORTH CAROLINA
OFFICE OF THE TREASURER JANET COWELL, TREASURER
Data Analytics
Project Inventory
As of March 2013, the following represent the catalogue of in-progress or proposed data analytics projects in the agency. The scope of the projects ranges from data collection, aggregation and presentation, to full exploratory and/or confirmatory data analysis projects.
It is especially worth noting that data collection and use must be on a timely basis. This is especially true for projects directly impacting the Department’s Investment Management Division. In the risk universe, a lag in developing a worthwhile product from the data collected erodes its usefulness. A decision made on outdated or aged reports from time sensitive data could be made worse than if the information didn’t exist at all.
The Department of State Treasurer has an enormous wealth of raw data that could be profitably used to drive management decisions and improve departmental effectiveness and efficiency. Specific projects to this end are listed below. The three data analytics and two risk management positions we are requesting authorization to create and fund would play key roles in each of these projects. Without these additional resources, much of this rich wealth of data may go untapped.
Office of State Treasurer
NC Economic Inventory
There are many free sources of public economic information, such as data from the US Bureau of Labor Statistics, the US Census Bureau, the National Bureau of Economic Research, and the US Dept. of Commerce, Bureau of Economic Analysis. The Department proposes to identify key indicators of economic activity, compile the data from the disparate sources, focusing on key data specifically about North Carolina and benchmark states, and re-present the inventory of economic indicators for internal and external use.
The automated collection and loading of data into this tool would save individuals a great deal of time in collecting this data from across the Internet. Presenting information on economic conditions is aligned with the Department’s goal on increasing fiscal literacy.
This project is in many ways similar to the County and Municipal Financial Dashboard released in 2010 – a web interface with aggregated data on financial indicators of all counties and municipalities in the State. Like the County and Municipal Financial Dashboard, this system would permit users to benchmark North Carolina economic indicators against other states and possibly perform regional comparisons. The aggregation of data would also provide a dataset for researchers to perform more advanced research such as trend analysis.
NORTH CAROLINA
OFFICE OF THE TREASURER JANET COWELL, TREASURER
Investment Management Division
Risk System
The Investment Management Division has partnered with SAS to develop a multi-asset class risk management and reporting system. This system, under construction, will be fed information daily on the State’s investment portfolio holdings and valuations and permit risk scenario analysis, assess
performance on a returns and risk factor basis, and model private investments to reflect underlying economic/financial risk factors.
When complete, this system will permit Investments personnel to understand the sensitivity of NC Retirement Systems assets and liabilities in different economic and financial environments. It will also allow staff to development different optimization and scenario techniques in the portfolio construction process to build more efficient portfolios.
Retirement Systems Division
Paid Leave Analysis
The Retirement Systems Division would like to conduct a focused analysis on the accrual of paid leave and its impact to the State’s pension systems.
Public employees may increase their retirement benefit by retaining paid leave time. Therefore, legislation and employer policy changes that affect the accrual of paid leave also impact the state’s pension liabilities. Over the past 20 years, the North Carolina General Assembly has passed a myriad of legislation increasing the amount of paid leave for public employees. This legislation is offered as a “no cost” substitute for pay raises. Through this project, RSD would like to measure the cost to the
Retirement Systems of annual leave, sick leave, and bonus leave to improve the foresight of our internal policy team and North Carolina legislators. A better understanding of the cost of the various types of leave will inform the evaluation of policy options.
This project will focus on the following research questions:
1. What actuarially relevant demographic features are associated with the amount of unused leave and/or the ratio of unused leave to total service?
a. Are the same factors associated with mortality correlated with the amount of sick leave available at retirement?
i. Since retirement benefits are paid as life annuities, an individual who lives for many years after retirement is more costly to the Retirement System than an individual who dies soon after retirement. We would like to know if those who
NORTH CAROLINA
OFFICE OF THE TREASURER JANET COWELL, TREASURER
live longer also use less of their sick leave during employment. If so, the cost of sick leave is greater than anticipated.
b. Is final salary or the growth rate of salary correlated with the amount of unused leave and/or the ratio of unused leave to total service?
2. On average, how much does a day of unused annual leave, sick leave, and bonus leave cost the Retirement Systems for individuals in different demographic classes?
3. How does unused leave differ across state agencies and local employers?
a. Are these differences likely to be the result of differential policies or differences in employee characteristics?
i. Some policies that influence paid leave accrual and use are determined at the agency or employer level. In the local system, the required employer
contribution rate may need to be adjusted to take into account any significant differences in policy.
b. Are there any outlying agencies or employers?
i. Leave policies must meet certain criteria specified by statutes. The Retirement Systems Divisions may need to investigate the policies of outlying agencies and employers.
4. On average, how much does a day of unused annual leave, sick leave, and bonus leave cost the Retirement Systems for employees at different agencies and employers?
5. How did the following policy changes impact the amount of unused leave?
a. In 1993, a law was passed allowing State employees to convert annual leave to sick leave.
i. Prior to this change, employees forfeited any available leave in excess of 30 days at the end of each year.
b. In 2001, the cap on the number of sick days that could be used for credit at retirement was removed.
i. Before the cap was removed, maximum amount of sick leave credit for a retiree with 30 years of service amounted to a 2.7% increase in their retirement benefit. After the cap was removed, a retiree with 30 years of service could accrue enough sick leave to increase their benefit by 4.7%.
c. In 2002, 2003, and 2005 the NC General Assembly passed legislation to provide additional annual paid leave (“bonus leave”) to state employees.
i. For an individual with a final salary of more than $60,000, this bonus leave is worth at least $1,000 per year in retirement benefits.
6. How much have the above policies cost the Retirement Systems and how much do we expect them to cost in the future?
7. Based on the impact of past policies, what is the expected cost of the 2012 bonus leave legislation?
a. To what extent do individuals use bonus leave as a substitute for annual leave and sick leave?
NORTH CAROLINA
OFFICE OF THE TREASURER JANET COWELL, TREASURER
Plan Design Policy
RSD seeks to utilize data analytics to assess how individuals respond to the incentives created by our plan design and whether these incentives meet the current human resources goals of state and local agencies.
For example, there has been a lot of discussion surrounding North Carolina’s difficulties with hiring experienced teachers and with turnover. Studies have shown that retirement benefits are especially important to teachers. If we were able to negotiate an agreement with the Department of Public
Instruction for data sharing, the Retirement System could play an important part in resolving some of the difficulties faced by NC public schools.
Data analytics should play a very important role in determining appropriate plan design changes to meet agency goals. Even with our current data limitations, we have the capabilities to determine how our current plan design is differentially impacting workers by gender, salary group, agency, age, etc. This sort of analysis may reveal some surprising information about possible biases that are not congruent with the Systems’ goals.
Communication Impact
RSD seeks to utilize analytics to measure the effectiveness of communications with active employees and retirees and make improvements if necessary. For example, identifying who is likely to respond
positively to certain types of communication regarding the Supplemental Plans.
Valuation Accuracy
Data analytics can be used to improve the accuracy of experience studies, leading to more actuarially fair reduction factors, more reliable liability predictions, and more complete valuations. This can inform future funding policy and ensure financial sustainability. Improvements in data accuracy and availability can lead to vastly different actuarial results.
Fraud, Waste and Abuse
RSD seeks to actively participate in the NC FACTS project and identify potential areas of fraud and waste surrounding the State’s retirement systems. Goals of the agency’s participation in this project are under development.
State Health Plan
Medical Utilization Cost
The State Health Plan has a fiduciary responsibility to the residents of North Carolina to monitor
utilization (visits per 1,000 members) and costs (per member per month) to help strategically plan for the future as well as identify trends that may require intervention. While this information is currently
NORTH CAROLINA
OFFICE OF THE TREASURER JANET COWELL, TREASURER
delay is receiving the data. Having this information available on demand will help end users work more efficiently by having the information readily accessible.
Using data from the Plan’s third-party administrator, the State Health Plan is developing a dashboard to identify the visits per 1,000 member utilization and per member per month cost for the major categories so information will be available in a timely manner. These are two measures that are consistently needed by various staff members for various purposes.
This project will provide end users with 4 individual web based dashboards and one overall dashboard: Emergency Room Utilization and Costs, Inpatient Utilization and Costs, Professional Utilization and Costs, Outpatient Utilization and Costs. Each dashboard will have several drill down options available (subgroup, age group, etc.) The overall dashboard will allow the end user to go to any of the individual dashboards.
Specialty Drugs
With the increase in specialty drug spend through the medical benefit it is necessary to manage SHP cost and identify the top specialty drugs that are causing the increase in medical spend. This information concerning specialty drug spend is used to validate rebates, monitor trend costs by using analytics to make the case if policy changes are necessary to control this increase in spend.
Using data from the Plan’s pharmacy benefit provider, this project uses a dashboard to identify the top 10 specialty drugs by cost, utilization and place of service when received through the medical benefit; with quarterly details for all specialty drugs using J-Codes. This project will provide end users with 3 individual web based dashboards: top 10 specialty drugs by cost, top 10 specialty drugs by utilization, and top 10 specialty drugs by place of service.
State and Local Government Finance Division
Self-Service Financial Analysis
The SLGFD collects volumes of financial information on over 1,000 units of government through vehicles such as annual audit review and the Annual Financial Information Report (AFIR). This
information is used by a number of different entities such as the US Census Bureau, UNC-CH School of Government, the North Carolina General Assembly, and other State agencies for a number of reasons, including the establishment of funding formulas.
Currently, the agency communicates fiscal health to the entities, but all reports and data visualizations need to be compiled manually. The collection of financial data into a warehouse with self-service analysis tools provided through the agency’s website would be a valuable tool units could use for their own monitoring. It could also provide capacities for interim transmittals of data, providing more up-to-date data for analysis and reducing the data entry load of the agency staff. Tools of this nature would also
NORTH CAROLINA
OFFICE OF THE TREASURER JANET COWELL, TREASURER
be valuable in assisting with the monitoring of the cash and investment data for each of the 1,250 reporting units each year.
Debt Data Analysis
Each year, over 350 debt applications are processed by the division for more than $5 billion annually. At the same time, local governments and public authorities have over $28 billion in debt outstanding. Today, most of the data surrounding a debt issuance is collected and analyzed manually. Analytics technology can improve the processing time of applications. It would be much easier to determine the impact of current debt on debt service requirements of the unit, evaluate their repayment of proposed debt, evaluate their resources and ability to repay debt and determine what amounts of debt the units can authorize and repay without a loss of financial flexibility.
Unclaimed Property Division
Escheat Fund Valuation
Beginning in 2012, the Unclaimed Property Division partnered with the Institute for Advanced Analytics at North Carolina State University to conduct an assessment of the State’s Escheat Fund. A team of graduate students in the Master of Science in Analytics program have been conducting research on property and claim data focused on five questions:
1) Using all known information regarding past claims, property reported, investment performance, and any other data available on the fund, what is the necessary fund reserve required to pay all future obligations? What if the rate of return increased by 10%, 25%?
2) What are the characteristics of property that is likely to be claimed – are factors like last known location of owner, property classification, years held or value of property significant indicators of whether or not a property is going to be claimed?
3) If recent trends of principal balance withdrawal continue, what is the rate at which the necessary fund reserve will begin to be depleted? How sensitive are the projections to changes in policy – in other words, how might a 5% per year increase in principal withdrawals impact the time horizon for fund reserve deficits and potential depletion?
4) How is principal depletion and investment performance impacting the fund earnings available for transfer to SEAA? How much investment earning has been lost? What is the optimal fund reserve to meet SEAA funding goals?
5) How can modification of property holder reporting requirements, such as increasing “aggregate” to $100 and dormancy periods, such as reducing the “catch all” dormancy period 3 years (NCGS 116B-53c(16)), impact valuation?
This information provide insight into the necessary reserve fund balance required in the Escheat Fund and how modifications with the program’s rules and statutes can impact the fund valuation.
NORTH CAROLINA
OFFICE OF THE TREASURER JANET COWELL, TREASURER
Outreach
Using information on past claims, unclaimed property, and other sources of demographics information, the agency can utilize data analytics for better management of its Outreach Programs. This project would focus on the key question: How do we maximize our resources in trying to return property back to its rightful owner?
By analyzing our data, the agency will be better informed as to the profile/demographics in which we should focus our outreach efforts. For example, an area of high concentration of unclaimed property could be the focus of outreach events, public service announcements, speaking engagements to educate that area. In a similar fashion, targeted communication and outreach could be done on owner profile groups. The agency should be able to both increase the return of unclaimed property to rightful owners, and do so in a more timely and cost-efficient manner.
Property Claims
The Unclaimed Property Division is statutorily obligated to approve or deny property claims within 90 days. Utilizing data analytics techniques, the division would like to project the resources needed to ensure that the obligation is met, even with a fluctuating volume of claims. The division would also like to apply quantitative methods to identifying process delays in the normal course of receiving and reviewing claims for approval and denial. Understanding process delays and trends or patterns in claim variability will permit better resource planning and provide lead time to permit “staffing up” for any influxes to ensure processing obligations are met.
Understanding the claims process in more detail will also enable more effective applications of process improvement efforts such as more informative instructions to the claimant, additional training or tools for our staff, etc. In total, this effort will improve the citizen experience when claiming property and allow more efficient processing within mandated guidelines.
Receipts and Audit
Through effective analysis of holder and property data, the agency can maximize the use of resources to ensure compliance with NCGS 116B. Data analytics may determine if certain industries are over or under reporting properties, or if certain common types of property are over or under reported. With this
information, the agency would be able to identify educational needs and increase compliance efforts. It can also identify opportunities to partner with industry leaders to recommend educational needs for their members and perform cost-benefit analyses on the value of audits. This type of data-driven management will permit the agency to increase receipts and reduce the cost of compliance enforcement.