Troubles With BBS CPI - II

Flaws in CPI calculation May prolong the bad situation.

Bangladesh Bureau of Statistics (BBS) provides this information about calculation of CPI.BBS prepares food and non food CPI for three groups :National ,Urban and Rural. CPIs for the three groups are classified further into twelve sections which include food,beverages, alcohol,housing,clothing transportation, recreation etc among other things.

Later two consumer baskets-- one for urban and another for rural---are derived from Household Income And Expenditure Survey(HIES) 2016-17,comprising of 383 items prepared for computation of CPI.An item's weight is prepared from its percentage share in the total expenditure on all items faced by a consumer.

[credit: BBS]

For data collection, BBS collects price data from 154 markets across the country. Among 154 markets, 90 markets are from urban and the rest are from the rural areas. Three price quotes for each item are collected from each market. Prices of 127 food items and 256 non-food items are collected. Another important thing to note is that rural and urban price data are collected on monthly basis. But data from Dhaka and Chattogram City Corporation areas are collected in every week.

Now some issues I would like to raise here for CPI calculation:

1. BBS could easily introduce two / four more groups(North-South or East,West,North,South) into the existing three groups of National ,Urban and Rural level. Average income and spending are different for different regions. Even the informal sector wage is different in these regions. Differences of income opportunities and consumption spending of these regions should be reflected in the CPI.
2. The 383 items do not reflect which one is tradable and which one is not. External sector has leverage on tradable items( i.e. food grain,cooking oil,fabrics etc). Changes in international prices and exchange rate variation influence prices of these commodities. On the other hand, nontradable items (i.e. housing, medicare,education,transportation, recreation etc) are often influenced by internal situation. This should be reflected in official reporting of inflation data. It will further dispel any policy ambiguity on inflation.
3. Data collection from 154 markets is not well balanced. Urban bias is more pronounced.Urban data are more recent than rural data. Urban consumption spending is different from rural consumption spending. CPI should remove this bias by allowing more rural markets into data collection process.
4. Jevons method is used for calculating CPI in Bangladesh. The other two formulae are Carli index and Detot index. But the problem with Jevons index is that it is deeply influenced by fall in prices. So when vegetables prices drop ,Jevons index is likely to produce a wider fall in overall CPI than as usual. In case of price fall,BBS can calculate CPI using Dutot index. Dutot index generally attaches more weight to expensive items and less weight to cheap items. So usual estimate of the CPI can be captured using Dutot index when there is a price fall ( See "Calculating Consumer Price Indices In Practice" by the IMF).
5. Calculating high level indices require individual item's weight from a base year,which is different and precedes the price reference period(p⁰). This weight is derived from Household Income And Expenditure Survey(HIES). BBS gets its weight from HIES 2016-17. Here comes the crucial thing : the more recent is the weight's base year ,the more pronounced is the variation in overall CPI from the variation of an item's CPI. So BBS needs to update the weights,based on more recent HIES survey. Though HIES survey is costly and time consuming, for the sake of greater macroeconomic stability regular HIES survey should be conducted in short intervals.

BBS inflation reporting is replete with bias and wider variation in some instances. A balanced,unbiased CPI reflecting just variation can provide true estimate of inflation. To attain that, more recent survey, representative data should be introduced.

Troubles With BBS CPI

Flaws in CPI calculation May prolong the bad situation.

February inflation data is out. Overall inflation declined to 9.32% from 9.94% in January. Food inflation for the first time reached to a single digit, shrinking to 9.27% from 10.72% in January. Prices of winter vegetables and import declaration of rices contributed greatly to the decline of overall prices. Govt has started procuring rice from various sources apart from the traditional importing countries. However, non-food inflation registered a slight increase in February, rose to 9.38% from 9.32% in January.

Many economists remain sceptics about the sustainability of drop in inflation in the coming months. They have valid reasons to remain so. What we are seeing now is the result of collective effort taken 3 months earlier in a bid to contain the inflation during the fasting month of Ramzan. Yet soybean oil, meat prices remain high. After Eid ,if this collective effort may dry up,a different scenario will prevail at the market. But I think the much hyped narrative of winter vegetables prices contributing to bringing down the price levels may turn out to be flawed perception upon closer scrutiny. Here comes the crucial question again: Are we calculating the CPI the correct way? Consumer Price Index computation assigns weights to individual items in case of calculation for multiple items. Here comes million dollar question: how do we assign the weight to individual items? Bangladesh Bureau of Statistics can answer that question to clarify lots of things. One simple approach may be based on importance. Now think of an individual consumer's monthly consumption basket. In this basket , carbs,cooking oil and proteins will get priority over vegetables. Goods that households have to buy (price inelastic demand) irrespective of price changes every month will not get the same priority as the winter vegetables enjoy. A consumer's spending behavior over months determines the priority. Now when vegetables prices drop, households definitely increase their consumption of vegetables but not to a great extent. A kilo of potato cost Tk 70 three months ago. With that price, one can now buy a kilo of potato as well as other vegetables in varied amounts. Without altering greatly one's monthly spendings on vegetables, one can increase his/ her consumption of vegetables. So vegetables' share in monthly spendings decides the importance consumer attaches to it. Then comes the question of spending behavior of different income groups and spendings in different regions. When BBS assigned the weight ,did it take into account these factors? I hope they did not randomly increase the weight of winter vegetables just to show the drop in inflation. If BBS did not randomly increase the weight of winter vegetables, then how did we construe that prices of winter vegetables bring down inflation when it occupies the lower rung in the consumption basket? This whole narrative of winter vegetables bringing down inflation is highly misleading. Just look at the formula of the CPI for multiple items:

CPI = (∑ _i=1ⁿ CPI_i x weight_i)/(∑_i=1ⁿweight_i)

Now think of just two items in the market: winter vegetables and rest of the items. Now looking at the sudden drop in prices of winter vegetables ,if you assign more weights to winter vegetables' CPI ,you will get higher drop in overall inflation. It is simple arithmetic. In this case , assignment of the weight is based on purely biased consideration. I just wanted to see winter vegetables is playing a role in overall inflation ,so I just raised the weights! Voilà! It is pretty similar to the incident of including laptop prices( a typical household here buys a computer/laptop once in every 10/12 years!) in consumer basket. BBS should come up with explanation how this weight on vegetables is determined. This whole thing also captures seasonal variation in the CPI. Now another crucial question arises: should seasonal variation in CPI be deemed important for long term trend? Three and half months later we are going to have Bakri Eid. Usually, meat prices remain low for two weeks after Bakri Eid. Based on this information ,if CPI registers a lower trend in inflation ,should we conclude lower meat prices bring down inflation?

BBS in the past involved in data engineering. Though this is an apolitical govt, some economic policies at the beginning indicated that it is following foot steps of its predecessor in some cases. So I am taking the inflation data produced by BBS with a dollop of salt. I hope and believe BBS has discarded past habit of data engineering. Only correct CPI calculation yields correct inflation data. And the right inflation data helps us setting the right interest rate. With flawed inflation data and interest rate, our economic recovery will only be delayed.