Ghana's inflation figures: the devil is in the details

At the time Dr. Bawumiah delivered his lecture dubbed “the true state of our economy” all who have in the past bemoaned the lack of quality in political debates thought it was going to usher in a period when substance would matter over dross. However, judging from the comments made by operatives of the ruling NDC and their sympathisers, this was but a forlorn hope. Whilst some NDC sympathisers have claimed that Bawumiah is no financial whiz kid, others have claimed he is only a technocrat but not an academic. I would have thought that if anyone is qualified to discuss the state of the economy it would be the technocrat rather than the academic with no professional experience.

Dr. Bawumiah made a number of points in his lecture, but the one which appeared to have stirred the most controversy is the claim that the statistically reported single-digit inflation is not aligned with other key indicators. The lecture has drawn reactions from the Ministry of Finance and the Ghana Statistical Service. The Ministry of Finance's statement was more of a political response. In parts, it sought to school Dr. Bawumiah on how inflation is calculated. It also accused him of intellectual dishonesty in the apparent mixing up of growth and debt figures from the new and old GDP series. The GSS, on their part, felt their integrity had been impugned and wanted to point out the fact that they subscribe to internationally acceptable principles and ethics in producing quality statistics. Instead of taking time to educate the public, though, the emotion-laden statement has rather placed the GSS in the midst of a political debate.

Now that the inflation debate appears to be subsiding, this article has set out to examine the issues involved. Its aim is to answer these three questions: how the rate of inflation is measured; what could cause a divergence between the official statistic and actual experience; and what could be done to ensure that the official statistic reflects the reality on the ground. Before anyone faults me for not citing my sources, I hasten to add that the article has drawn on data and publications by the Ghana Statistical Service (GSS) and the UK's Office of National Statistics (ONS).

How is rate of inflation calculated?
There are two bases for calculating inflation: the Consumer Prices Index (CPI) and the Retail Prices Index (RPI). Both measure the change in the general level of prices charged for goods and services bought for the purpose of household consumption. They are compiled from the average change in prices across a wide range of consumer purchases, which is obtained by recording the prices of a sample of products from month to month using a sample of markets throughout the country. The main difference between the CPI and the RPI is that weightings are applied to the price changes to compile the CPI whilst the RPI compilation uses no weighting. Thus the RPI measures the changes in the total price of the entire inflation basket. The inflation figures announced monthly are the percentage changes in the indices relative to the same month of the previous year. Ghana, like many other countries, uses the Consumer Prices Index (CPI) as the basis for calculating the rate of inflation.

The Ghana Statistical Service captures the prices (relative to base year 2002) of 242 items, obtained from 40 markets throughout Ghana. The 242 items constitute the 'inflation basket' and is meant to reflect the spending pattern of a 'typical' Ghanaian household. The items are classified into twelve 'Consumption by Purpose' (COICOP) item groups, which with their relative weightings, are as follows:

• Food and Non-alcoholic Beverages (44.91%);
• Clothing and Footwear (11.29%);
• Alcoholic Beverages, Tobacco and Narcotics (2.23%);
• Housing, Water, Electricity, Gas and other Fuels (6.98%);
• Furnishings, Household Equipment and Routine Maintenance of the House (7.83%);
• Health (4.33%);
• Transport (6.21%);
• Communications (0.31%);
• Recreation and Culture (3.04%);
• Education (1.60%);
• Hotels, Cafes and Restaurants (8.28%); and
• Miscellaneous Goods and Services (2.99%).

These item groups in turn have sub-groups. The 'weightings' assigned to the item groups and sub-groups are fixed to reflect their importance in household spending. Both the items in the inflation basket and the weightings are obtained through household surveys and from national accounts data.

The 'Food and Non-alcoholic Beverage' Division is also referred to as the 'Food' Category. By having the combined weighting of 44.91%, the monthly reported inflation rate in Ghana is largely driven by the prices of food items. Within the 'Food' Category, inflation is mainly influenced by the prices of: vegetables including tuber vegetables (weighting = 12.46%); fish (weighting = 10.24%); bread and cereals (weighting = 7.97%); and meat (weighting = 4.07%).

The remaining eleven item groups (i.e. apart from 'Food and Non-Alcoholic Beverages') constitute the 'Non-Food' Category with a combined weighting of 55.09%. It can be seen from the weightings that inflation within the 'Non-Food' Category is influenced mainly by the prices of: Clothing and Footwear; Hotels, Cafes and Restaurants; Furnishings and Household Equipment; Housing, Water, Electricity and Fuels; and Transport. Of relative less importance are price changes in the Health and Education item groups with weightings of 4.33% and 1.60%, respectively.

Hitherto, what has not been adequately reported in the media is the fact that the headline monthly rate of inflation is not uniform across the country. The GSS also publishes monthly regional inflation figures, which in recent months show that the Central Region is consistently top of the inflation chart (12.1% in April 2012). The lowest inflation figures are obtained in the Upper East and West Regions (6.3% in April 2012). There is also a variation between urban and rural areas with the rural inflation figures being consistently less than urban ones.

How could the calculated rate of inflation differ from people's experience?
Dr. Bawumiah posed a pertinent question that so far those who are criticising him have not adequately answered. He asked that: if single-digit inflation is not leading to falling cost of living; falling bank lending rates; is not leading to creation of jobs; and is not leading to stabilised exchange rates, then what is it for? Let us examine the factors that could cause the statistically reported inflation figures to differ from actual experience. Even though Dr. Bawumiah was categorical in not doubting the integrity of the GSS, he was of the view that the difference between actual experience and what should have been the benefits of single-digit inflation may be due to: measurement issues and what is being measured; as well as lack of adequate resources and independence for the GSS.

First of all, the CPI from which the rate of inflation is calculated is a 'geometric mean' statistic. There is no household in Ghana that will mirror exactly the spending pattern covering all 242 items in the inflation basket in the same proportions suggested by the weightings. Hence individual household experiences are not going to be exactly as indicated by the headline inflation figure. However, if the national figures suggest a slow down in the rate at which general price levels are increasing, individual households should be experiencing the same trend in their spending budgets.

As stated above, the CPI is compiled from a sample of prices, of a sample of consumer items, in a sample of markets, at particular dates of the month. Biases can therefore be introduced in each of the sampling exercise if adequate care is not taken to make the samples representative. In all cases, increasing the sample size would lead to increasing convergence between the official statistics and actual experience. I could not find any information on the number of prices captured monthly by the GSS, but as an example, the ONS in the UK calculates the consumer prices index (CPI) rate of inflation by using 180,000 price quotations every month, covering 700 goods and services, from 150 areas across the UK.

The weightings assigned to the item groups are meant to be proxies for the quantities (in price terms) of the items in the inflation basket consumed by households. If they deviate significantly from actual spending patterns, the calculated statistic would differ from reality. In the UK the weightings are much more evenly spread over the 12 item groups than in Ghana. For instance, 'Food and Non-Alcoholic Beverages' has a weighting of 11.8 % compared to Ghana's 44.91%. This may be due to the differences in spending patterns between the two countries. However, the 44.91% weighting is markedly different from the 30% stated in the 5th Ghana Living Standards Survey (GLSS5) Report. Looking at the weightings, the Vegetables sub-group is of relatively higher importance (responsible for about 1/8th of the CPI changes) than say Education, which is responsible for less than 1/60th of the CPI changes. It would be highly educative for the GSS to explain the basis of the weightings to the Ghanaian public.

Another potential cause of divergence between statistically reported rate of inflation and actual experience is the coverage given to each 'Consumption by Purpose' item group. A group which contains sub-groups with widely varying price movements has to be covered much more widely than one in which price movements are generally aligned. Again in the case of Ghana, whilst the prices of 59 items are captured in the 'Clothing and Footwear' item group; 43 in the 'Housing, Water, Electricity and Fuels' item group; only the prices of 2 items are captured in the 'Education' item group, which has 5 sub-groups. Is the coverage given to Education wide enough to capture all price movements for the different levels in both private and public establishments? Only the GSS can answer this question.

Lastly, to ensure that the inflation basket is representative of people's spending patterns, it has to be kept up-to-date in terms of: the markets and shops people visit; the goods and services that people buy; and the amounts people spend on them. Thus it is not so simple to just say cement is not in the inflation basket, unless household spending on cement is insignificant. In the UK the basket and the weightings are reviewed annually during which time items could be dropped from the basket as new products are admitted to reflect changing spending patterns. For instance in the last review in March 2012, Tablet computers were added, as they represented a significant and growing market, whiles charges for developing and printing colour films were removed, as the popularity of digital cameras meant that fewer and fewer people were using analogue cameras. It is not possible to tell, from the publications on the GSS website, how often the items in the inflation basket and the weightings are updated. It does appear from the historical series that these have remained unchanged for a number of years. If this is so, then as more years elapsed since the last update, divergence (if there is any) between statistically reported rate of inflation and the reality on the ground could become even wider. This is where political independence is needed. If the GSS has to receive its expenditure through the Ministry of Finance, which Finance Minister would be willing to fund the revision of the inflation basket and weightings, which could potentially result in an upward revision in the rate of inflation? It would be like a turkey voting for Christmas.

What improvements could be introduced?
From the foregoing, there are several factors that could affect the accuracy of the inflation figures. For the official rate of inflation to keep up with the spending patterns of Ghanaian households, the most important improvement would be to annually update the inflation basket and weightings. We could also think about increasing the sample sizes of: the number of items in the inflation basket; the number of prices captured; and the number of markets and locations where prices are collected. This of course would require additional resources for the GSS, which brings us back to the call made by Dr. Bawumiah in his recent lecture.

Meanwhile, whilst waiting for the day that the GSS would be adequately resourced, they should proactively put out more information about how they calculate the monthly rate of inflation. They should also consider calculating the Retail Prices Index (RPI) based rate of inflation alongside the CPI rate from the same data they collect each month. The RPI, which charts the total price of the inflation basket without the effect of the relative weightings, would help to determine if the relative weightings currently applied are introducing undue distortions. In the UK, the rate of inflation based on the RPI is reported alongside the CPI rate of inflation. These two measures together are used to inform government policies and the level of welfare benefits.

Despite the sharp political differences that currently prevail in Ghana, it should be possible for all to join the call for the strengthening of our institutions. In this quest, it is also not misplaced to ask for more resources and independence.

May we ever march forward in the building of stronger institutions!

Dr Yaw Ohemeng
Manchester, UK