This article was written based on a request from UQBS marketing by the Financial Management Magazine. An excerpt was subsequently published in the February 2020 magazine release. Below is the full copy of the article

What are the main benefits of implementing RPA in investment banking and fund management?

RPA as in using data science and applied AI to optimize and automate processes has the advantages of maximizing human resources towards towards creative , problem solving tasks, rather than menial repititive everyday tasks.

Investment banking has become a very competitive industry such that bankers need to be one step ahead of the game to pre-empt the needs of clients and provide advisory services before the client requests it.

Examples of investment banking preempting needs of clients are by identifying short selling activity, shareholder activists, sub-optimal capital structures, and potential targets for M&As.

The above-mentioned tasks are traditionally performed by analysts using spreadsheets to analyze balance sheet and accounting statement data of each firm. Loading this data onto spreadsheets and analyzing it is largely a manual process. RPA can be used to evaluate a portfolio of companies to obtain answers quickly about which clients to approach so more time an effort can be spent focusing on client needs.

A practical example is that data analytics are used to observe historically instances where firms of a certain industry or vintage have made decisions to have a bond and equity offering and identify similar firms that exhibit similar characteristics at present so they can be approached by investment bankers offering pre-emptive advisory services.

For funds management, especially for quantitative funds, RPA is used to detect whether erroneous returns data has been entered into trading system as this can impact decisions made by portfolio managers or risk managent systems. The process of data cleaning has often been a manual process this RPA helps to detect which data points are erroneous outliers that need to be handled appropriately. Another example is that sometimes trades are not executed properly or are not filled. This can be to do with awkward timing of when orders are being made, illiquid assets, exotic instruments, country/industry of the investment, etc. RPA can be used to flag and detect when such orders are being made to inform portfolio managers to exercise extra diligence and caution when making a trade, rather than following standard processes only to have to come back to it later to correct or fix any errors that have transpired.

How should banks and fund management firms of different sizes and workloads plan RPA implementation?

Smaller banks or fund managers would hire consultants to evaluate what the manual processes are and how many humans are being used at present and what the potential savings are by optimizing the process.

Larger banks and fund managers build internal consulting teams to perform such work. Some asset managers have internal teams that perform this role so well that they begin to use these teams to generate additional consulting income to the firm.

Which stakeholders need to be involved and who should lead the project – finance or IT?

Successful RPA teams usually run via IT providing consulting services to finance. This IT leads the project, understands the scope of works and needs of the finance Dept. iT executed and seeks to deliver to the satisfaction of finance.

How can CFOs and their teams navigate the RPA vendor market?

The most effective way to engage vendors if looking for a long term relationship is to break up a large internal project into a smaller chunks. Provide two vendors with the same tasks and a similar timeframe and evaluate their performance. This way you can benchmark them against each other.

Although it seems like an initial double spend on resources , it can save substantial time and effort as some vendors tend to over promise and under deliver. Thus, if you over commit to a single firm it is easy to be dragged along and continuously being billed with little or no results.

Which finance processes are most and least suited to RPA?

Payments, billing , travel claims or any activity that happens frequently are most suited for RPA. For example, often travel claims are frequently entered into the system and require approvals. RPA can be used to detect where a travel expense is most likely to be true and accurate and immediately approve as.oppsed to wait for a human to check and verify its veracity .

Low frequency finance activity such as expenses for purchase of large capital equipment (not maintainence) are not ideal.

Generally activities that happen frequently and produce alot of historical data where one can understand the processes to automate or train a mathematical model to evaluate a processes accuracy or completion are most suited.

What common problems do finance leaders encounter during and after RPA implementation?

Adoption and trust is usually the biggest issue. Humans are reluctant to trust an automated system and will focus on the rare moments were the RPA fails rather than on when it succeeds.

RPA also needs a well defined scope of works with reasonable expectations set out by the team delivering the solution to a client.

When is RPA not a good solution for the finance function?

RPA can be a poor solution where insufficient data exists or if the data is highly unbalanced. This is where the activity is infrequent such as large capital purchases.

What’s the biggest myth or misconception about RPA?

Biggest myth is that it destroys jobs or leads to more efforts in correcting or addressing any errors made. The reality is that humans can be spent it more value added activity and the errors that an RPA system should be few rather than frequent. If errors happen too often, this is because it has not been implemented properly or the scope of works is not well defined.