Can we measure infrastructure investments more accurately?

Published on Friday, 14 July 2017

Published by Kaushal Joshi, Janine Elora Lazatin and John Paul Flaminiano on Friday, 14 July 2017

The non-infrastructure share of GFCF (GG) in Pakistan is 63%.
The non-infrastructure share of GFCF (GG) in Pakistan is 63%.

When quantifying Asia’s infrastructure investment needs, figuring out how to accurately measure such investments is crucial. Failing to do so can grossly misrepresent needs, and further delay badly needed upgrades to the region’s roads, power lines, water supply networks, and other facilities.

In the absence of a standard indicator to measure public infrastructure investment, one of the most commonly used measures is Gross Fixed Capital Formation for General Government, or GFCF (GG).

But, is it precise enough? Is there a better way?

GFCF (GG) measures investments in fixed assets by the government, including national and subnational governments and social security funds. This indicator is readily available from the IMF’s Investment and Capital Stock Database for 170 countries from 1960 to 2015, which makes it suitable to analyze long-term public investment trends.

However, there are two major pitfalls of using this indicator GFCF (GG) to estimate general government infrastructure investment.

  ADB definition of infrastructure more restrictive than GFCF (GG)

First, it includes government investments in fixed assets in all economic sectors, but ADB’s Meeting Asia’s Infrastructure Needs report defines infrastructure as including only transport, energy, water supply and sanitation, and telecommunications. Other economic sectors such as manufacturing, and social infrastructure—education, health, etc.—are excluded. 

Second, GFCF (GG) reflects the total investment made by a government on fixed assets of all types, which also includes investments that are not necessarily infrastructure-related. Thus GFCF (GG) includes all types of residential and non-residential buildings, civil engineering works, machinery and equipment, weapons systems, cultivated biological resources, and intellectual property products.

ADB’s definition only features investments in civil engineering works and infrastructure related to non-residential buildings and machinery and equipment.

We can illustrate our argument by using disaggregated data from national accounts statistics from Fiji and Pakistan to “decompose” GFCF (GG) so we can see how much of it is actually infrastructure (see Figure 1). The data show that in Fiji, non-infrastructure investment accounts for about 23% of GFCF (GG), while the non-infrastructure share of GFCF (GG) in Pakistan is 63%.

While on one side GFCF (GG) may overestimate general government investment in infrastructure, on the other side we are still missing information about investments undertaken by the state-owned enterprises (SOEs). This indicator may therefore underestimate public infrastructure investment if SOEs in an economy are heavily acquiring infrastructure assets through investments.

However, at this point, we have no evidence of the direction of the bias.

Standardized framework, disaggregated data

So, is GFCF (GG) the best way to measure public infrastructure investment? We believe it is not.

To come up with a more accurate measure, disaggregated data used to estimate GFCF (GG) in national accounts can be extended to provide a standardized framework for satellite tabulations from national accounts.

If sufficient information was available, this method would result in data that complies with common statistical standards and allows more precise comparisons across regions, economies, sectors, and institutions.

One challenge to implementing this approach is lack of standardized classification mapping from economic sectors in national accounts to infrastructure sectors considered above.

For instance, national statistics offices often classify investments in roads as GCFF (GG) under public works (or public administration) instead of transport. In that case, GFCF (GG) classification by economic sectors in national accounts would underestimate the actual investment level.

  We need complete measure of infrastructure investment for an entire economy

The other challenge would be to segregate the non-infrastructure related investments in non-residential structures, civil engineering structures and in machinery and equipment.

Improving infrastructure investment measures through analyzing disaggregated GFCF data would require additional work by national accounts statisticians to compile infrastructure investment data by economic sector, by asset type, and by institutional sector.

The feasibility of doing this would largely depend on the availability and level of details on acquired assets in the basic sources of data, such as government budget statements and financial statements of SOEs and private corporations. 

Aside from being able to correctly classify assets into infrastructure investments, there is also the bigger task of standardizing the methodology across economies.

Despite the mammoth task ahead, coming up with satellite tabulations for infrastructure investments using standardized data from gross fixed capital formation will allow us to obtain a complete measure of infrastructure investment for the entire economy.

This way, we will have a much clearer understanding of actual investment flows in each period, be able to segregate infrastructure investment from non-infrastructure investment. A more precise and accurate measure of infrastructure investment can better inform policy-making to help realize our #LetsBuildAsia goals.