What is Programme 2232?
Programme 2232, “The International Fellowship for Outstanding Researchers Program”, aims to support qualified researchers with leading scientific and/or technological achievements and international work experience to conduct their research in leading university, industrial or public institutions to contribute in areas that are strategic value to Turkey.
About the Project
In this proposed research project, we have two key objectives.
Our first objective is shaped around constructing a NLP algorithm to establish a wealth of new data sources that moves financial institutions into the realm of big data in Turkey. Specifically, regular scrutiny of banks’ balance sheets has become vital following the financial crises and hence there are some key variables such as the evaluation of prepositioned collateral by financial institutions and detailed information on their balance sheets, as well as an increasing set of micro-market datasets ranging from mortgages, over news sentiments to access transaction-level data in over-the-counter markets. However, the manual inspection of hundreds or thousands of firms records’ is not only inefficient, but also quite challenging.
Through NLP, we will analyze the key constitutions of annual reports by automatic analysis of words and phrases in financial commentary and business strategy sections in the financial statements and hence portray their level of opaqueness in financial disclosure.
Why is Information Disclosure Important?
Disclosure is the act of providing information to the market, while transparency arises only if the information is reliable and can be appropriately interpreted and used by the market. It is this concept of transparency that underpins effective market discipline. In order for markets to operate effectively investors need access to information about financial institutions’ risk profiles. Given the balance sheet risk, if sufficient transparency and monitoring by investors impose incentives on companies to hold less risky positions, then companies that disclose more information should choose less risky activities. In other words, investors or debt holders may exercise a direct market discipline, allowing a reduction in the company’s risk. A company’s annual risk report intends to reduce the information asymmetry between the bank and its stakeholders. Companies’, banks’ and insurers’ disclosures provide an important channel through which relevant information can be obtained. Market participants can then use this information to influence the behavior of firms in which they invest, allowing risks to be assessed and priced accurately; a mechanism known as market discipline.
Our second objective is to apply this transformative methodology of ML to time series setting and present a modeling framework to forecast and nowcast the main macroeconomic indicators in Turkey on a medium-term horizon of two years. Most empirical policy work focuses on causal inference, and we instead prefer to develop our understanding in the predictive inference. More importantly, our analysis will provide additional and more timely information beyond what we already know from latest macroeconomic data such as growth, inflation, credit and unemployment. The results of the project are expected to benefit the CBRT’s decisions, especially for managing the systemic risk in the financial markets.