Wednesday, October 5, 2016

Explaining Inflation…Again

I had some input into an article the Malay Mail published last month:

Is there really a gap between the official inflation rate and reality? Well, yes and no

KUALA LUMPUR, Sept 5 — As Malaysia’s urbanites begin to feel the bite of soaring living costs, there is growing suspicion that the official inflation rate does not quite reflect the economic reality….

…So is there a gap between real consumer experience and official data? The explanation itself is quite technical but in short, it’s a yes and no.

On one hand, economists worldwide have long decried the method used to measure inflation—the consumer price index (CPI) — saying it is far from reliable.

Domestically, there have been debates about whether or not the CPI model accurately depicts the reality on the ground....

They only used some of my reply (because, as usual, it was long-winded), so I thought I might take the liberty of publishing my remarks in full:

1. Is there is a gap between the inflation rate and true inflation? If not then why is there a perception that there is?

I don’t believe there is much of a gap, and if there is one, the CPI would tend to overstate rather than understate, underlying inflation. This is a view shared by specialist economists in this area, and is largely due to qualitative improvements in the goods we buy. For example, there have been enormous advancement in terms of mobile phone hardware, in terms of capacity and capability, yet prices have barely increased for phones aimed at various market segments. The same thing goes for things like cars, TVs, and so on. So while the CPI is reasonably good at capturing quantitative changes in prices, it isn’t very good on the qualitative side, which has improved considerably over the years.

Also, in terms of whether you can trust the CPI numbers, research in this area (such as MIT’s Billion Prices Project) shows that the CPI does reasonably capture the rate of price increases across many different goods and services.

The question remains as to why the perception of inflation is higher than official data, or what we can glean from overall actual price increases (as in the MIT project). This is not something unique to Malaysia either, and is a global phenomenon. Surveys of inflation expectations in the United States for example, consistently show that people think inflation is 2-3 times higher than it actually is (roughly 5% versus 2% as of the latest reading).

The main reason for this is because perceptions of inflation are largely driven by the things we buy every day, versus the things we buy less frequently. This point is important because inflation in recent times has been largely driven by food (including prices at restaurants) and petrol, whereas durable goods prices (which we buy infrequently) have been relatively stable or even declining. This is borne out by the sub-indexes of the Malaysian CPI, where durable and semi-durable goods prices have been flat or decreasing, whereas nondurable goods and services prices have been consistently increasing.

A second reason is that people also think the CPI is supposed to reflect increases in the cost of living, which it actually does not. The CPI only captures the prices of goods and services we consume, but not all the prices related to living expenses, and in particular home ownership. The housing component of the CPI only includes rent and not overall house prices. The technical reason for this is that rent measures the consumption of housing services, while buying a house is an asset purchase, which provides a flow of housing services across time and not just at the point of purchase. While rental costs have been increasing, it has done so much, much more slowly than the prices of houses, which leads to a divergence between consumption prices and cost of living expenses.

Thirdly, some people have confused high consumer prices with high inflation. These are not directly related – it’s possible to have high inflation but a low cost of living (Indonesia) or very low inflation and a high cost of living (Japan). A high cost of living, and low affordability is just that and no more, and says little about the rate of inflation.

2. So how exactly is one suppose to analyse the CPI or the inflation rate?

To address some of the major criticisms of the main CPI index, Malaysia’s Department of Statistics (DOSM) actually publishes alternative indexes, including rural and urban indexes, state level indexes, and an index covering the low income group (for those earning less than RM3000 a month). Outside the main report, new indexes covering those earning less than RM1500 and RM1000 are also available online for analysis.

In addition, even the components of the main index can yield valuable information, and conform closer to people’s perception of inflation. For example, the latest CPI report published for July 2016 shows meat prices have increased 7.0%, vegetables 6.3% and seafood 6.0%, even as the overall index increased by just 1.6%. The biggest reason why we have low overall “inflation” for July despite the jump in food prices, is due to petrol prices being lower this year than last year, and because of the impact of GST implementation in April last year dying away, as far as the growth calculation is concerned.

Another alternative way of looking at inflationary pressure is to focus on the month-on-month change (0.3% in July, or equivalent to an annualised 3.7%) instead of the year-on year percentage change, which can be subject to “base effects”. A high base last year (as happened with the mostly one-time increase in prices due to GST), automatically lowers the year on year growth rate, even if the pace of price increases hasn’t changed. This is a purely mathematical artefact of how growth is calculated.

Lastly, I’d like to draw attention to the new Core inflation index, which was introduced by DOSM recently. It excludes not just prices of seasonally volatile items (such as vegetables) but also price-controlled items (such as petrol and diesel). This gives a much better picture of underlying inflation than the main CPI index does, while directly addressing the critics,

3. Do you think our CPI model is reliable?  Or do you think it can be improved and how?

Overall the CPI does a decent job of portraying inflation, but as noted above, the Core CPI is probably a more important measure. Best practice in developed countries is to report both headline and Core CPI, as well as both year on year and month on month growth figures. DOSM has done a good job in extending the methodology to cover areas of interest (e.g. state level CPI and CPI at different income levels), but I would like to see CPIs at the city level, which might give important insights into price developments beyond the state and/or urban/rural differences.

4. CPI and inflation is key to determining wages but wages have somewhat stagnate [sic]. Is this a perception or is it true? If so how do you reconcile the gap between the benign CPI and wage stagnation?

Actually, the relationship between inflation and wages is two-way. Inflation is definitely a factor in determining wages, but wages are also a key determinant of inflation. Labour costs can be a large driver of price increases, particularly in labour intensive industries. Good examples of this are in education, healthcare and in restaurants. As Malaysia goes up the income ladder, prices of services will rise in tandem with wages. Outside of food price inflation, which is more due to global changes in demand and supply, we are already seeing that happen as services prices have increased at roughly twice the rate of overall inflation.

On the larger question of wage stagnation, it’s largely not true with one important caveat. Aggregate wage increases have more than kept up with overall inflation, but we have segments or industries where this hasn’t happened. Conversely, there are certain industries and certain income levels where wage increases have far outstripped inflation. So while on average Malaysian wages have increased in after adjusting for inflation, some Malaysians have seen no wage increases at all.

A good example of what I mean here is graduate employment. Malaysia is seeing the fruits of vastly increasing access to tertiary education, while at the same time the number of people in each successive graduating class is also getting larger. But that substantially increases the supply of graduate labour relative to its demand, and plays a role in supressing wage hikes for graduates.

Looking across income classes instead, the minimum wage implementation in 2013 had a very big impact on wages for the bottom 40% of the wage distribution, but did very little for the middle 40%, where real income growth has been relatively slow.

So the picture here is really one of yes and no, and depends a great deal on which segment of society and which industry you’re talking about.


  1. Hi Hisham,

    You've said there is an "index covering the low income group (for those earning less than RM3000 a month)". Whats that index?


    1. @Group 8

      The basic problem is that the CPI averages out expenditure across all the households surveyed. But different income levels on spend more (or less) of particular item groups. If the rate of price increases are different across items, than different income levels will actually be experiencing different rates of inflation. Hence the need for indexes covering households of different income levels.

      For example, lower income households spend a bigger part of their budget on food compared to higher income households. But food inflation is higher on than most of the rest of the CPI. Therefore, the lower down the income scale, the higher the rate of inflation you should be experiencing.

      I actually just found out to day that there's a technical term for this, called Engel's Law:

  2. Always have this question in mind, finally getting a platform to ask: who decide and how do they decide the weightage of each item in CPI?

    1. @Fung

      You decide, metaphorically speaking.

      DOS conducts a household expenditure survey every couple of years or so (I think they've gone to an annual schedule now). They get the survey households to record what they spend on and how much.

      The survey data is then aggregated, and that is used to determine what goes into the consumption basket and what the weights are.

      It's not arbitrary.

  3. Hishamh, your latest article in The Star referred. You mentioned that corner shops almost extinct due to hypermarkets, well I think they are being replace by 7/11 (and their copy-cats).

  4. Everyone is seeing price of every thing is up after GST, so does the blind and deaf. Surely, the data provided to nation is incorrect.

    1. @anon 10.43

      If you look at the details of the CPI, you'll find that individual prices are fairly accurate.

      The problem is that people don't bother counting in the prices that have fallen, particularly petrol. RON95 is down 10% from last year.