The Asian Voice

Can inflation be better measured with big data?: Jakarta Post contributors

The writers say robust information about inflation is crucial especially for a developing country to allow better remedies and to prevent further severity.

People trade at a traditional market in Jakarta on Feb 7, 2022. PHOTO: AFP

JAKARTA (THE JAKARTA POST/ASIA NEWS NETWORK) - Inflation is one of the most important anchor indicators, particularly in emerging economies where consumption is predominantly spending on food.

Consequently, increasing prices might directly oppress those in the lower-income bracket forcing them to tighten their belts and crimp spending on other household spending resulting in slower economic growth.

A recent press release by Statistics Indonesia (BPS) seemed to reflect a steady inflation rate in Indonesia. We experienced a low inflation rate of 0.66 per cent in March 2022 (month-to-month), after deflation of 0.02 per cent a month earlier. The core inflation, measured without taking food and energy costs into account, was even lower at 0.30 per cent in March 2022 or almost constant, compared with 0.31 per cent in February 2022.

The reality check might be somewhat different. Thanks to price apps such as Indomaret and Segari, we can observe approximate increases in prices in some commodities during February-March, for instance sugar (up by about Rp 1,500/kilogram), bread (Rp 1,000/pack), flour (Rp 1,000/kg), eggs (Rp 3,000 for 10 eggs), and chicken (Rp 3,000/kg). These were additional to the infamous increases in cooking oil and cooking gas prices. Blue gas has increased three times in just two months from Rp 160,000 to Rp 180,000 (S$15.20 to S$17.10) and eventually reached Rp 200,000 per 12 kg, whereas the Bimoli cooking oil price rose from Rp 27,000 to Rp 47,000 for 2 litres.

Is there any bottleneck in the calculation? Why do the official statistics report more stable prices than what we might observe in the market?

To recall, the inflation rate is measured based on the change in the consumer price index (CPI) over time. It indicates the overall cost of goods and services purchased by society to portray the changes in the cost of living.

Gregory Mankiw in Principles of Economics lays out five steps to calculate inflation. The relevant statistics agency needs to firstly determine the basket of goods being calculated to best mimic the overall cost of living standard. The basket might be different across regions.

For example, some regions might be weighted-up with a greater volume of rice whereas others with flour. Once the basket has been decided, the basket cost is calculated by multiplying the price and quantity to obtain the total spending on consumer items. The agency then decides the base year to calculate the inflation rate as the percentage change in the CPI over the years using the same determined base year.

The challenges might concern the selection of the basket index. Two indexes are often used: the Laspeyres and Paasche indexes. In principle, both methods aim at picturing the ratio of spending in the current year compared with the base year.

The main difference lies in the quantities being used: the Laspeyres index uses the quantity in the base year, whereas the Paasche index uses quantities in the current period. BPS adopts the Modified-Laspayres method to portray the price index.

With this approach, the agency might prioritise monitoring solely prices without necessarily paying attention to the availability of stock. This means that the quantity of goods available at a particular price level is not being observed.

To illustrate, the government introduced a price intervention for cooking oil about Rp 14,000/litre in March before abolishing the intervention. However, the quantity was extremely limited at that price. People lined up for cooking oil at the fixed price or to buy it at a more expensive price elsewhere.

Nonetheless, implying the Lespeyres method, official statistics will only report the intervention price of Rp 14,000 in the basket of goods whereas the quantity was fixed based on the living survey held in earlier years.

Consequently, while theoretically the Lespeyres index tends to overstate inflation, in Indonesia the opposite could happen. The index might understate inflation as the administrative price fails to reflect the real market price for people at large.

This problem is not specific to Indonesia but also other countries using indexation to measure inflation. With the advent of computational power, new machine learning and scientific data mechanisms, it is possible to complement the traditional process by optimisingg automatically generated quantity and price data via any point of sales (POS). These include data gathered from supermarkets or any e-commerce transaction via gadgets such as iPads, mobile phones or computers at home.

This information is processed through big data containing a massive volume of both structured and unstructured information so that voids concerning real-time quantity and price can be better tackled.

Some initiatives have already been introduced to mitigate these challenges by international organisations and central statistics agencies. The Economic, Social Council for Asia and the Pacific (UNESCAP) statistics division has been arguing in favour of these alternative sources being included in official statistics. In the United States, the Bureau of Labour Statistics has been working with alternative data sources such as using retail sellers' data, which is more timely and cost saving.

In Japan, Abe and Shinozaki on their 2018 study showed that alternative data on commodity prices from various comparison websites may provide greater accuracy than those of more traditional surveys. Some developing countries have also taken up this initiative.

Robust information about inflation is crucial especially for a developing country to allow better remedies and to prevent further severity. The IFG Progress study details that inflation might be transitory or persistent requiring corrective measures. Either way, it has a potential impact on labour market developments and economic growth.

As the famous management guru Peter Drucker once said, "If you can't measure it, you can't manage it." This is the best time to incorporate big data technology to help measure inflation.

  • Ibrahim Kholilul Rohman is senior economist at IFG Progress and lecturer on digital economics at the School of Economics and Business, University of Indonesia. Moinul Zaber is researcher at the United Nations University. The Jakarta Post is a member of The Straits Times media partner Asia News Network, an alliance of 23 news media organisations.

Join ST's Telegram channel and get the latest breaking news delivered to you.