Wednesday, March 20, 2019

The CPI and the Cost of Living: Cross Country Comparisons

This is hopelessly unscientific, but it's an interesting exercise, and is a useful starting point for understanding what goes into the cost of living. So what I have here are comparisons of the weights in the CPI for various countries in comparison with Malaysia, selected from a set of relatively large countries with roughly (±10%) the same level of PPP-adjusted GDP per capita (as per the CIA's World Factbook). The CPI weights are taken from national sources (Malaysia, Poland), Haver Analytics (Mexico), or the IMF (Greece, Hungary, Portugal, & Turkey), and grouped according to the UN's Classification of Individual Consumption According to Purpose (COICOP).

First, some big, big caveats:
  1. CPI weights are ratios to average household consumption expenditure, so don't take these as indications of differences in the cost of living between these countries. To measure that, we need the local currency costs, which aren't readily available. All the numbers in the graphs below are in percent of average expenditure;
  2. My critique of the CPI as an indicator of the cost of living still applies;
  3. Differences in the data could also be interpreted as differences in cultural preferences (alcohol consumption, or eating out for example), or regulatory or legal requirements (e.g. sin taxes), and not just from underlying differences in costs between countries;
Now on to the comparisons:

A quick summary:
  1. Food & non-alcoholic beverages: Malaysians in general spend more on food than anybody in our peer group. Rural Malaysians also spend relatively more, although this was expected, as food features more in the household budget for lower income families;
  2. Alcoholic Beverages & Tobacco: Either we don't smoke and drink so much, or sin taxes are too low;
  3. Clothing & Footwear: We don't spend so much on clothing either, or maybe four season countries see a much higher turnover in terms of clothing (though that doesn't explain Mexico);
  4. Housing, Water, Electricity, Gas, & Other Fuels: Owning and maintaining a home is expensive in Malaysia. We spend 250% more than the Portuguese, and it's very obviously an urban Malaysian phenomenon;
  5. Furnishings, Household Equipment, & Routine Household Maintenance: Home furnishings on the other hand, is relatively cheaper here (or we're better at bargaining);
  6. Health: Either health costs are dirt cheap in Malaysia, or we aren't spending as much as we should;
  7. Transport: Transport costs, surprisingly to me, is about the same across the whole sample;
  8. Communications: We spend relatively more on communications (too many Facebook peeps), though not by much;
  9. Recreation Services & Culture: We're about in the middle for recreation;
  10. Education: Education spending, like health, is on the very low side;
  11. Restaurants & Hotels: The same goes for restaurants and hotels (cheap foreign labour maybe?)
So what stands out? On the high side, we have food, housing, and communications. On the low side, alcohol & tobacco, clothing, furnishings, health, and education. The relative sameness of transportation across the countries in the sample, as I said, was a bit of a surprise.

From a policy perspective though, it's clear that food (mainly for rural Malaysians, though the expenditure share is high for all Malaysians) and housing (for urban folks) are the obvious starting points for further digging. Expenditure on communications is too low, and the differences against other countries too small, to have a big impact on household budgets (though every little bit helps). I suspect some of the differences in the expenditure weight on food comes mainly from differences in income levels between urban and rural. Another possibility is that Malaysians simply like our food, and we eat too much, which corroborates with our national obesity levels. Nevertheless, these are areas that are certainly worth investigating.


  1. Some thinking points...

    1. It's great to control income levels, especially for food, but is it possible to control for household size, and female lfpr

    2. Ditto above, but subsidy effects

    3. Would chain-linked cpi methodology cause some variation in cross country comparisons?

    4. Wouldn't economic structure matter? Agriculture productivity, infrastructure quality, tech concentration

    5. Related to above, food/fuel import dependency, trade exposure to China for manufactured goods?

    Wouldn't mind a collaborative effort if you're thinking of doing one, since my dept is looking at a whole lot of countries at once

    1. Hi Jason,

      1. You'd have to have the detailed survey data, and I don't know anybody who publishes that. Also, why female lfpr?

      2. Ditto above, but on prices

      3. Since the purpose of the post is comparing current expenditure shares, no

      4. Since I'm controlling for national income, not really

      5. The only oil exporter in the list is Mexico. And everybody is exposed to China for manufactured goods

      6. Depends on what you guys are doing it for

  2. 1. Would hypothesise that higher female lfpr would result in higher household's consumption basket to be skewed towards discretionary spending, especially for dual income households (but again, probably need to control for household size)

    Re: 4. & 5.: again another hypotheses where agri-based economies (not necessarily lower income) may skew the results towards discretionary spending, same examples also apply for different structures

    6. Well it's just a thought seeing that improving our understanding of inflation and prices would help in our analysis of our countries and also rationalising policy reactions to these conditions/structures.

  3. Salam tuan Nurhisham Hussein,

    Congratulation to you on your new appointment as Astro alternate director.