On Saturday, I found that the report from Bank of England. This report is about economic analysis in central banks with Big data such as social network services. It is good not only for economic researchers, but also business personnels to consider how Big data should be used. So I would like to consider it based on this report for a while.
Before considering usage of Big data, I would like to define “Big data”. Big data is data sets that are granular, in real time basis and non-numeric data as well as numeric one. These data are completely opposite in nature compared with data which are currently analyzed in Central banks. Because such data are usually “aggregated, periodical and numeric”. One of the examples is financial statements of companies. Big data are different from such data. For example Twitter are generated by individuals in real time. These are usually text, images and video. Then the questions come.
1. Can we build up macro economic models based on big data?
Central banks are responsible for the stability of the financial system in the country. Is it possible for central banks to collect data of each loan from private banks and assess credit risk of each, then confirm financial stability as a whole country? It can be applied to private companies, too. Is it possible that the company collect data of each customer, forecast the amount of purchase by each customer and predict the revenue of the company next fiscal year? Big data may enable us to do so even though it takes time.
2. Is the method used “theory based” or “data driven”?
Even though they cannot be clearly distinguished in practice, these are two approaches to analyze Big data in economic analysis. Someone puts importance to economic theories. Let us call it “theory based”. Others take another approach of “Let the data speak for themselves”. We may call it “data driven”. Their opinions are sometimes against each other even though they analyze the same data. So we should have well-balanced approach between them.
3. Should we change the processes to make business decisions?
Big data comes to us in a real time basis. But our decision making process in organizations is usually periodical. For example, board of directors meetings and executive committees in companies are generally held on a monthly basis. Should they be held more flexibly in a timely manner based on outputs from analysis of Big data, rather than periodical one? The bigger companies become, the more difficult it is to change the process in practice.
FRB in the US is currently wondering when they should raise the interest rate of the US. Chairwoman of FRB has been always saying “It is based on economic data“. But I am not sure she cares about data (conversations) on social networking services in the US. What do you think?