The shifts from SST to GST, and back to SST were controversial. There were debates on its effects on inflation and living costs. While the GST introduction led to a one-time rise in price level and that, I think, has not been widely disputed, the effect of SST reintroduction on price level has been more contentious. Pro-government camp would would claim that SST reintroduction did bring prices down, while the opposition would disagree, and citing various examples where prices went up (both real and made-up instances).
Did SST bring prices down?
The answer, on aggregate, is yes. Here is the main chart central to my argument. This chart shows core price level.
In April 2015 when the GST was introduced, prices went up by 1.19% month-on-month (after adjusting for seasonality). And in June-September 2018 when the SST was reintroduced, prices fell by 1.32% month-on-month (also after adjusting for seasonality).
Now, the long explanation.
Core prices instead headline prices
There are various factors affecting prices. It can be difficult to extract and isolate each factor out. But the official core inflation series does a relatively good job filtering out various factors, most notably items with volatile prices affected by seasonality. This cancels the noise (enough) that we can see price effects of tax regime changes.
Using core prices as our starting point, did the SST bring prices down?
Month-on-month versus year-on-year
The answer, again is yes… from month-on-month perspective.
This section is a bit dry and digresses into discussion on measurement. You might want to skip this as skipping it would not hurt the “yes” explanation by too much. If you are not skipping, let us move on with the agenda.
Month-on-month is a better way to observe things than year-on-year. This is because there are multiple significant supply-induced for the past few years. For instance, because of the SST which was reintroduced in September 2018, we would have to wait until September 2019 until its effects on prices disappear from year-on-year perspective. These changes make year-on-year unreliable as a change measurement. If we insist on using year-on-year as a measurement for policy purpose, we will risk making the wrong call based on a massive structural break, never mind the 12-month recognition lag.
Year-on-year does control for seasonality, unlike month-on-month and that is its strength. But it is not great at handling (structural) breaks in series. And as far as consumer prices are concerned, Malaysia has experienced too many breaks at least since 2015 when the GST was implemented. The last big break was in February 2019 when petrol and diesel prices were capped at its current ceiling. I expect one or two more major breaks in the next 12 months.
This is why there is a need to move from year-on-year to seasonally-adjusted month-on-month (or quarter-on-quarter whenever relevant) measurement. Seasonally-adjusted month-on-month addresses problems of structural breaks and seasonality, which makes it better than year-on-year by a mile.
Additionally, changes in a series could be seen immediately through month-on-month. This significantly removes the problem with recognition and policy lags.
Our familiarity with year-on-year and stubbornness to move away from it is part of the reason why too many people panicked over “deflation” earlier this year, when in fact, it was just a mathematical artefact arising from a massive structural break, or two, or three.
Effect of shift to GST in April 2015
Now that is out of the way, we can start directly discuss about how GST changed the price level.
The raw month-to-month price change in April 2015 was 1.28% (below is just the month-to-month change for the chart above).
But how do we know whether the 1.28% was fully due to GST?
That is a difficult question, really. But because it is core prices, significant amount of items susceptible to large volatile changes are out of the picture. Food items and fuel are out. Yet, there is still a problem. Our problem is that the core CPI is not seasonally-adjusted (as far as I understand it). In order to control for seasonality, we need to look at March-April change in other years and use that as a correction factor.
Here, we get into another problem, the official core series does not go to far back. Publicly, core series begins in 2015.
Nevertheless while keeping that in mind, April price change in 2016, 2018 and 2019 were either 0.08%, or 0.09% (in April 2017, prices rose 0.26%. I do not know why, and I am too tired to find out. So I am going to close my eyes and consider it as a outlier and pray hard. Please do not shoot me). To control for seasonality, we take the 1.28% and subtract 0.09% (the April average change in 2016, 2018 and 2019) from it. Through this, we can claim that the April price change due to GST—as far as core prices and the seasonality I have accounted for concerned—is 1.19%.
In short, GST quite possibly raised core price level by 1.19% month-on-month. Yes, GST did raise price level.
Effect of shift to SST in June-September 2018
Now, this part is not so straightforward because the transition from GST to SST lasted for 4 months. In 2015, there was an immediate transition from SST in March to GST in April. But in 2018, the GST was effectively abolished in June 2018 and was replaced with a 3-month tax-free period. The SST only came in September 2018.
As a result, direct comparison between the SST-GST shift (April 2015 core CPI), and GST-SST shift (September 2018 core CPI) could not be made. There are at least two reasons:
- First, in June 2018 when the GST was abolished, core price level dropped 1.43%. And it was a drop from GST regime to no tax period. This number does not help us answer our original question, which is whether SST brought prices down.
- Second, when the SST was reintroduced in September, core price level rose by 0.60%. This also does not help answer the question.
So, how could we make it comparable?
This chart shows the problem, and the adjustment required from price level perspective to make a fair comparison (blue is actual, red is adjusted).
The adjustment is: calculate the difference between May 2018 core price level (GST prices) with September price level (SST price level), while ignoring the free-tax period completely. Do this and we would get a price drop of 0.59%.
This is how it looks like in month-on-month changes (blue is actual, red is adjusted).
However, just like the SST-GST shift, we need to control for seasonality. And the average May-September change for 2015, 2016 and 2017 is 0.73%. The variance is not that big: for transparency, it was 0.72%, 0.79% and 0.68% rise for 2015, 2016 and 2017 respectively.
Controlling for seasonality, that means the GST to SST shirt brought core price level down 1.32% (that is -0.59% plus -0.73%).
Conclusion: SST brought core price level down
As a summary, after accounting from seasonality:
- SST-GST shift raised core price level by 1.19%
- GST-SST shift cut core price level by 1.32%
Some comment on the results: I had expected the SST-GST and the GST-SST shifts to bring prices up and down by about the percentage point. Right now, there is a difference because I think there are still some important factors that have not been controlled. It is possible that one of them could be forex rate.
Caveat and other business
There is a important clarification here. These results do not mean core prices in September 2018 when the SST was reintroduced were lower than core prices in May 2015. Between May 2015 and September 2018, headline and core prices rose for a variety of reasons. Rather, the two changes in 2015 and 2018 were one-time changes, or structural break in core price series. You could see this from the price level charts above.
This brings us to inflation. The remarkable thing is, at least from the naked eyes, inflation (in the sense of general rise in prices) remains largely the same throughout the changes. Another way to say is that, the slopes of the lines (GST and SST) are about the same. What happened was a shift in price level, which is what this whole post shows.
And finally, I want to show you this chart.
The black-shaded area was caused by GST (specifically, GST at 6% minus GST at 4% – GST at 4% has been cited as the equivalent of SST around 2014-2015). If GST were never introduced, core price level would likely have be lower by that shaded area.