[2777] Rebasing, revision and GDP-ratio targets

From time to time, economic statistics get revised. Usually statisticians require a lot of time to compile data and in that mad rush, certain data could left out first and included only later when everybody gets a chance to reflect. There is nothing structural about the revision. It is just about errors, corrections and business as usual.

Other times, the revisions are more structural. Some are structural only because of definition change like what happened with the concept “external debt” last year. Others include very deep changes. An example of that is the GDP rebasing exercise and it affects policy targets.

The Malaysian GDP gets rebased once every five years and the exercise consists of two parts: rebasing and revision.

The rebasing itself is simply a manipulation of index but the more significant part of the exercise is the revision that include/exclude of new/old sectors. Strictly speaking, the change in the composition of the GDP is not rebasing but instead, it is a structural revision. It is really the revision that makes rebasing such a big deal.

The revision is a problem for any policy with GDP-ratio targets as it can make such targets quickly irrelevant. Since Malaysia structurally revises its GDP once every five years (for instance, from 2010 to 2014, the GDP base year was 2005. For 2015 till 2019, the base is 2010), any GDP-related target formulated in 2013 for instance could become problematic in 2015 when a new GDP series is used.

Here are two examples.

First is the 55%-to-GDP debt limit that the Malaysian government maintains. Notwithstanding the off-the-budget spending criticism as well as the fact that the limit itself is a paper tiger and assigned arbitrarily, the government promises to keep its debt below 55% of GDP. Previously, a lot of people were worried that the government would breach the limit. Not so much now and this is largely because of the revision.

As you can see, the old GDP series (with the 2005 base) has the government cutting it close but under the 2010 GDP series, there is a lot of space still for fiddling around:

Effect of GDP revision on Malaysian debt limit

The implication? It gives the government more room to borrow just because the GDP statistics have been revised upward while allowing the government to keep to its words.

Another example is the fiscal balance of the federal government. You can see, the Malaysian fiscal deficit ratio is slightly lower under 2010 GDP series compared to the 2005 series.

Effect of GDP revision on fiscal balance

The ratio changes are not trivial from policy perspective.

In the case of deficit, previously thought to be a severe policy under one GDP series might not be so severe under the other after all. For instance, the federal government recently revised its deficit target from 3.0% to 3.2%. But 3.2% deficit under the 2005 GDP series is harder to achieve than it is under the 2010 GDP series. If the government sticks with the 3.2% target after the rebasing/revision, then the government could have higher absolute deficit and actually borrow more than it would have if there was no rebasing/revision exercise.

To put it simply, the goal post moves and it becomes larger.

This is part of the reason why I prefer to target deficit on government revenue instead of  on GDP.

I suppose the other way to correct for this is to tighten those targets every time there is a rebasing exercise.

And there are other policies beside fiscal that look at GDP-ratio too.

I think the revision would become less of an issue if it is done every year. The problem with doing it once every five years is the sudden jump, which can throw a lot of targets into questions. Policymakers make targets simply on incomplete and dated data. In fact, any target made based on the status quo would be softer than it looks like.

A yearly revision would solve that and make any GDP-ratio target more robust.

By Hafiz Noor Shams

For more about me, please read this.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.