Documenting Income Inequality: 2012 Update

I posted the data on household income over a year ago, but the 2012 household income survey summary results have since been released, and I’ve had quite a few queries about this issue lately.

[As always, you can click on any of the graphs to view a larger version. Note that all the data presented here is inflation adjusted (2000=100); please refer to the sources linked to at the bottom of this post for the raw data]

So this post serves as an update to my original post, incorporating the latest data up to 2012 (in RM per month):

Here we see overall income gains have accelerated since 2009, while overall inequality has decreased:

…except for Malaysian Indians, where the Gini coefficient has risen from 0.424 to 0.443. Over the medium term, income inequality appears to be on the downtrend (again, except for Malaysian Indians), although progress is painfully slow.

A look at the income strata data shows where Malaysian Indian households are diverging (ratio of mean top and middle incomes against low income households):

Here’s a closer look at the “Indian” problem:

High income Malaysian Indian households are showing substantial income gains over both middle and low income groups. While Malaysian Indian incomes are close to the average for Malaysia as a whole, the distribution of that income is skewed towards the top. Anecdotal evidence suggests that this skewing of income is present even in the top 20%.

From a different perspective (ratios to Malaysian average):

High and middle income Indian households have gained in the last 3 years (the top more than the middle), but poor Indian households have seen a continuous decline relative to Malaysian averages since 1997. I should note however, that the ratios are still all above unity i.e. higher than the Malaysian average.

Which leads me to the comparison with the Bumi position, which is especially appropriate given the pending announcement by the PM this weekend. And here we can see why (RM per month; ratio to Malaysian averages):

The substantial progress of Bumi incomes relative to the Malaysian average seen since 2002 has effectively stopped, except for the bottom 40% of households, although even here progress has slowed. This is contrary to my expectations, and just as it was hard to explain the reasons for the gains seen between 2002-2009, it’s been equally hard to figure out why Bumi income growth has since lagged, especially at the top of the distribution.

For contrast, here’s what’s happening to Chinese households (RM per month):

As noted in my earlier post, incomes for Chinese households in the top 20% and middle 40% were effectively flat or declining from 2002 to 2009. Since then however, they have advanced across the board.

To underscore the point, here’s how the incomes of Chinese households have compared to the Malaysian average (ratio to Malaysian averages):

We see the same picture here of income generation stronger than the Malaysian average.

So the story that emerges is that although overall inequality and inequality within ethnic communities is being slowly reduced (except as noted, among Malaysian Indians), intra-ethnic disparities have generally begun to widen again (ratio against Malaysian averages):

Not unexpectedly given these developments, the rural-urban income gap has stopped closing as well:

These are all troubling developments.

I continue to be flummoxed attempting to explain these movements and trends in incomes across time as no single factor appears to be responsible, or at least in a way that is amenable to statistical testing. There’s certainly more than a few candidates for causality, as a read of the comments section of my previous post shows (the comments are well worth your time reading). But at this stage, they continue to be no more than hypotheses.

This is a terribly important question to answer – not knowing the manner and direction of causality means policy making based on supposition and not on evidence. And if you can’t quantify what you’re doing, there’s no objective yardstick for measuring success or failure.

Technical Notes:

1. Data on household incomes from the Economic Planning Unit
2. Summary results from 2012 HIS from DOS
3. Consumer price index data from DOS (spliced series rebased to 2000=100)