The overall risk management of a firm is largely based on statistical measures. This is because most of the models used for managing and mitigating risks are based on advanced statistics. These statistical models are based on gathering large volumes of complex data, rigorously processing it via running complex statistical operations, and using the resultant metrics to make reasonable predictions about the future.

Now, it needs to be understood that all the predictions are based on past data. Hence, it can be said that data is the lifeblood that runs through all these statistical models. Hence, it is very important to ensure that only high-quality data is used as input in these models. The failure to do so will lead to inaccurate results which undermine the credibility of the entire process of risk management.

Common Issues with Data Quality

Since the risk management field has been in existence for a long time, there are several issues with data quality that risk managers have already seen. The company should set up a process to ensure that the data used in the process does not have these known issues:

1. When millions of data points are entered into the system, a few data errors are expected and can be considered to be the norm. However, if there are a lot of erroneous data points, then the accuracy of the entire data comes into question.

   It is important for the organization to have some checks and balances to ensure that the data being used in the model is correct. This can be done by comparing entered data to the source data or by comparing processed data with unprocessed data.

2. In many cases, the entered data may be accurate. However, it may not be complete. Important data points which have a material impact on the calculation of risk metrics may not be present in the data point. The presence of such data points makes the entire process unreliable. Companies need to have a system in place to ensure that the completeness of data points.

3. The data being used in the model should not be inconsistent. This means that the one data point in the model should not contradict another data point in the same model. There are various types of consistency checks which can be used to ensure that the correct type of data is used.

   For instance, record level consistency is when two different sets of data values within the same record match. Similarly, cross-record level consistency is also checked. This is when the values of different sets of data behave in a similar manner even if they are in two different records.

   Finally, there is temporal level consistency which looks at two sets of data points within the same record. However, both sets are from different points in time. If a data set matches all the above criteria, it is said to be consistent.

4. The data being used in the model should be current. This means that it should belong to a time period that has a bearing on the results which will be used in the risk management process. If the data used is of an incorrect or irrelevant time period, then the results are also incorrect or irrelevant.

Data Validation and Data Quality Inspection

Many large organizations have data quality governance programs in place. The objective of these programs is to ensure that the quality of data being used is validated and quality tested. All major organizations across the world spend time deciding and streamlining all the major processes which are required to ensure the smooth flow of good quality data to the models.

A well-crafted data governance policy is fundamental to the risk management program of any company. This is the reason why these procedures are often outlined within the risk management policy of the company.

Data validation is a one-time step that is used to ensure that the data transferred between the source and the risk management system is correct. This is done to ensure that no errors have taken place during transmission.

In contrast, data quality validation is an ongoing process. The objective of this process is to ensure that the quality of data being used in the model matches some predefined quality requirements. Also, if any data points are found to be incomplete or incorrect, the quality validation team is in charge of correcting the data and bringing it up to an acceptable level.

The purpose of the data inspection team is to identify data quality issues at the earliest. It is the job of the data quality inspection team to clearly define what will be considered to be an unacceptable data point, who will be in charge of rectifying this data, under what situations, and using which methods.

Lastly, it also needs to be understood that data governance is not actually the same thing as master data management. Data governance is on a macro level and takes a bird’s eye view of data quality issues. On the other hand, master data management is engaged in the day-to-day management of data.

The bottom line is that data quality is absolutely instrumental in the successful functioning of any risk management department. Hence, it is the duty of any organization to ensure that good quality data is being used in the process. Otherwise the proverbial “garbage in garbage out” model will become a reality and the results obtained after so much hard work will not be applicable.

Related Articles

View All Articles