Advanced Analytics for accurately forecasting currency demand for the largest South-Asian Central Bank
Case Study

Advanced Analytics for accurately forecasting currency demand for the largest South-Asian Central Bank


The client is the largest central bank in South-East Asia responsible for liquidity management, currency issuable, financial supervision, foreign currency reserves, exchange rate policy, and other national level economic policies.


Banking & Financial Services


With the accelerated changing economic landscape, most of the algorithms were failing to deliver on the specific needs of the central bank. Extraneous shock events such as elections, demonetization, etc. were simply adding fuel to fire, not to talk of the significant developments in the applications of electronic technologies in the retail payments areas across several such countries.

The bank wanted to have an at a glance accurate forecast of the demand for banknotes and coins for different denominations. They also wanted to know the numbers for replacements or retirements of banknotes as well as the effect of different shock events, such as elections, on currency demand.


Datamatics examined alternative techniques that do not solely rely on structural relationships between money and other nominal or real variables. Hence in addition to estimating currency demand functions, Datamatics seriously also explored econometric time series based algorithmic interventions and implemented advanced analytics solutions.

Datamatics helped the bank to develop a user-friendly solution powered by R and Dot Net framework. It used the latest Econometric Time-Series Forecast modelling and the analysis of 10 years of data for Banknote and Coin requirement. The data was extracted from 4 different sources – Withdrawal, Retirement, Buffer Stock, and Currency-in-Circulation.

The solution used concurrent, multiple model engagement for building precision and accuracy. The solution had the ability to handle shock events seamlessly and effectively. Both long and short term estimations were head-and-shoulders above both in precision as well as in accuracy in comparison to the other competing and erstwhile estimates.


99% Precision Achieved

99.9% Precision Achieved

In forecasting requirement of bank notes and coins

100 Accuracy

100% Accuracy

In stock taking of currency notes that involved withdrawal, retirement, buffer stock, and currency-in-circulation

20% Reduction

20% Reduction

In the cost of printing/ minting the currency in the first year itself