By W. D. Lakshman –
The Department of Census and Statistics (DCS), the official statistical agency, has recently uploaded into its website, an undated report on “Rebasing National Accounts Estimates.” DCS uploads into its website similar reports regularly, at least a few in a month. Only a very few of them would usually come under public scrutiny as closely as the above report has over the last few weeks. Indeed this “Re-basing” report came into public notice, not through the DCS website, but through the political statement made by Deputy Minister Dr. Harsha de Silva around the DSC report and brought to public attention by daily news telecasts. As may be many others, I also was made to look up the report in the DCS website only after this political statement of the Deputy Minister.
As most countries in the world, Sri Lanka too follows the United Nations’ System of National Accounts (SNA) in the preparation of its national accounting statistics. The SNA, first introduced in the early 1990s, has its latest revised version published in 2008. The aggregative economic measure known as the Gross Domestic (or National) Product (GDP or GNP), which is now perhaps a household word, is derived from a national accounting exercise. Those who are even superficially familiar with the subject know that the GDP is derived as the value of output less intermediate consumption or the sum total of values added produced by all productive agents in an economy. The computation of GDP can be made through three methods – product, expenditure and income methods. The SNA is a system of coherent, and integrated national accounting framework in which the statisticians try to estimate a country’s national product aggregates using these three methods simultaneously.
In no country in the world do statistical authorities have access to all necessary sources of information to estimate aggregates like the GDP with a 100 per cent accuracy. Making best possible use of limited available sources of information and sometimes undertaking statistical surveys to fill gaps in essential data sources, they make estimates of aggregates like the GDP as best as they could, working on various simplifying assumptions linking the available information to the variables forming part of the national aggregate concerned. Under these circumstances, the estimate of GDP for a particular year should not be taken as an absolutely correct measure of GDP but only as the best possible estimate which the officials in service can arrive at under prevailing data availability conditions.
In these circumstances, it is nothing but right that the methods and conventions used in the GDP computations are revised from time to time as there are various changes taking place in underlying conditions as time passes. The development of national accounting systems in Sri Lanka commenced in the early 1950s under severe limitations. Since then these systems have undergone change and revision on many subsequent occasions. The latest such revision is this change announced a few weeks ago. The change is described as one of “re-basing” from a 2002-base to a 2010-base. This base year change implies the following. In the national accounting data published so far, 2002-based price indices have been used to convert the domestic/ national product estimates “at current prices” into “constant price” estimates. The announced change is that the DCS would now on begin to use 2010-based price indices for this purpose. A GDP data series at constant prices is called a series of “real” GDP and are used to work out the widely used growth rate numbers.
The change in statistical practice announced by the DCS is, however, a broader and more comprehensive change than a mere “re-basing”. It involves a revision of the activity classification, together with an increase in the number of activities taken into account, in the national product computation. It is noted that 40 broad economic activities were covered in the national accounting exercise earlier, and in the new system, the coverage has been raised to 48 broad economic activities. In addition, the data sources and data collection methods in the GDP computation have been strengthened. Changes have been introduced in respect of price and volume indices used in the conversion of “current price” estimates to ”constant price” estimates although the DCS’s internet paper does not provide details about these changes in indices. This document also mentions a number of other changes introduced to the national accounting practices.
The DCS discusses the impact of the “re-basing” of national accounts on several significant economic variables. One set of these effects is summarised in Table 1. As should be expected, the new series of GDP estimates show higher values than the old series, as the number of economic activities coming into the GDP computation process has been raised in the re-basing process. As Table 1 shows, for all the years for which GDP numbers were re-computed, GDP in the new series is higher than in the old series by proportions varying between 5.2 per cent and 15.2 per cent. The lowest percentage difference recorded between the old and the new series for 2014 may be noted. Per capita GDP at current prices, stated in US dollar terms (last two columns in Table 1), also shows higher values in the new national product series than in the old one. The re-computation of GDP has also influenced sectoral composition (or the structure) of the Sri Lankan economy broken into agricultural, industrial and service sectors. The contribution of the agricultural and industrial sectors to the GDP in all the years from 2010 to 2014, according to the old series of GDP estimates, was higher than according to the new series. Accordingly, the new series shows a greater service sector bias in the economy than was shown in the old series. The new series also shows, over time, a gradual increase in the service sector dominance of the economy, whereas the old series has shown a gradual decline of service sector dominance during the 2010-14 period.
I have begun this article by referring to some political controversies which this “re-basing” article in the DCS website gave rise to over the last few weeks. None of the points discussed so far in this article about this DCS “re-basing” exercise was responsible for these controversies. The controversy that continues is about the implications of this statistical exercise of the DCS for growth rate estimates for the two years, 2013 and 2014.
The rate of growth of a country, despite its weaknesses as an economic measure, has come to be widely used as a dominant macro performance indicator. Several countries which exhibited continuous high growth rates – around 8-10 per cent per annum – for about 10 years in the recent past had succeeded in “economically taking off” and achieving perceptibly high levels of economic well-being for their people over a relatively short period as a single generation. Examples from among our Asian neighbours include countries like Japan, South Korea, Hong Kong, Singapore, Taiwan, Malaysia, People’s Republic of China and so on. The achievement of a high and sustained economic growth rate has been a national policy goal in Sri Lanka too at least since the end of the 1970s. The highest annual rate of growth we managed to achieve, however, was 8.2 per cent and until the beginning of the decade of 2010s, such high growth rate achievements could not be sustained for any extended period of time. Until the publication of this new series of GDP data, we believed that our rate of growth since year 2010 remained above 7 per cent excluding the year 2012 with its 6.3 per cent rate of growth. There were two years (2010 and 2011) during this period with 8 per cent or higher growth rates. According to the new GDP series, the 2012 growth rate – the lowest for these five years according to the old series – was the highest for the period at 9.1 per cent. The high growth rate claim of this period was not only a “statistical” phenomenon. There were also other strong elements of impressionistic evidence around us at that time making us believe that the growth rate numbers presented by statistical authorities were indeed realistic and clearly in the domain of acceptability.
The growth rate concept looks at the total production activity in a country over time in real terms, or by removing the impact of price inflation on the value of outputs produced through such productive activities. As explained, GDP or GNP is the statistical measure of this total output expressed in value added terms. Changes of GDP over time have two components, the change in real output and change in prices of the goods and services produced. The rate of economic growth over time is measured using GDP valued at “constant prices”. Any series of values expressed “at current prices” can be converted into values “at constant prices” by deflating the former by an appropriate price index. Prior to the development of the new series of GDP discussed in this article, the current price GDP was deflated by a series of 2002-based price indices to derive constant price GDP numbers. The new series of GDP at current prices, however, is deflated by a set of 2010-based price indices to derive the constant price estimates. This is what the DCS calls the re-basing of national accounts. Table 2 presents the impact of this re-basing exercise on the estimated annual rates of growth over the four years 2011-14.
While noting the significance attached to the “growth rate” by all political regimes, I have pointed out that prior to 2010 Sri Lanka had failed to attain and sustain a high rate of economic growth for an extended period of time. The political regime in power during the high growth era of 2010-14 was claiming the credit for facilitating such high growth rates during this period and for being able to maintain those high rates consistently for half a decade. Knowingly or unknowingly, the DCS has uploaded the results of its so-called “re-basing” exercise at a sensitive time of the run-up to a Parliamentary election. A DCS report on national accounting, which otherwise would have gone unnoticed except by a few professionals, has become a source of political debate.
Whatever it may be, what could be an economist’s objective view of the data sets presented in Table 2 above? The problematic numbers are those for 2013 and 2014. In any attempt to explain and understand the drop in the growth rate as between the old and the new series of GDP estimates, one has to look at current price GDP estimates and price indices used in the deflation of current price estimates. As already noted, the estimates of GDP at current prices for all five years 2010-14 are higher in the new series than in the old, but by different percentages. Particularly noteworthy is the 5.2 per cent difference in 2014 in contrast to 2-3 times larger differences recorded for the other four years. It is unfair on the part of the DCS to present a new series of statistics which has numbers so widely divergent from the old series without explaining at least some significant factors responsible for these wide divergences between the two sets of estimates. In deflating current price estimates to derive constant price estimates of GDP, the statistical authorities use different price indices. Again no information is provided about the price indices used, 2002-based and 2010-based. Because of the lack of required information to understand and explain, I would be hesitant to ditch the older series, simply because a newer series has been made available. The time at which this document was uploaded into the DCS website and the manner in which this document’s “unexplained downgrading” of 2013-14 growth rates was used by the government in power for its political benefit are further reasons for being cautious in the analytical use of this “re-basing” exercise.