Growth and Structure of the Economy

Abstract

The foundations of the present structure of the Ghanaian economy were laid between 1890 and 1910. This 20-year period witnessed an annual average growth of 1.8 per cent in GDP per capita according to estimated national income accounts for that period. Judged by the economic performance of developing countries at that time, such a growth rate was high and marked a significant improvement in living standards. As observed by Omaboe (1966, p. 18):

This was the period during which the export economy of the forest belt of the country was developed and transformed. Prior to this the country had a small export trade but this was based largely upon the collection of naturally-occurring forest produce such as palm fruits and kernels, kola nuts and wild rubber. These two decades saw the replacement of this export trade by the product of two major economic activities, gold-mining and cocoa-farming. They have dominated the economy of the country for more than half a century now and they have dictated the pace of economic growth and the present structure of the economy.

Appendix to Chapter 3: Ghana’s National Accounts Statistics

A considerable amount of the macroeconomic analysis in this book depends upon an interpretation of the national accounts statistics. It is therefore particularly important to recognise that Ghana’s GDP data were revised significantly in recent years with a new set of data released in 2010 giving an overall level 60 per cent higher than with the old set. This ‘upgrade’ of GDP data was based on the 1993 version of the United Nations System of National Accounts (UNSNA) while the previous series had been based on the 1968 UNSNA (GSS 2010a, b). The upgrade affects the estimation of a long series of GDP data (which cover a period both before and after the new series was adopted) and of ratios for a number of important series as a proportion of GDP—such as those for Public Expenditure to GDP, for Tax to GDP, for Imports and Exports to GDP, and for Official Development Assistance to GDP. We need to gratefully acknowledge the cooperation of the Ghana Statistical Service (GSS) in making available detailed data from the new national accounts series and in providing a basis for interpreting them. Of course, the World Bank’s World Development Indicators needed to be updated in order to fully accommodate the new data series. Background to some of the West African national income statistics is provided from the first chapter in an introductory economics textbook first published in the mid-1960s for which a second edition was published in 1987 (Livingstone et al. 1987), and from a chapter in a recent handbook on African economic development (Jerven 2015b).

A significant cluster of publications followed the release of Ghana’s new GDP data in 2010, not least being Jerven’s book entitled Poor Numbers (2012). There are also relevant articles in the African Statistical Journal (Jerven and Duncan 2012), the Review of Income and Wealth (Devarajan 2013; Jerven 2013a), African Affairs (Jerven 2013b), a special issue of the Canadian Journal of Development Studies (Jerven 2014a), and a United Nations University WIDER Working Paper (Jerven 2014b). The 2010 GDP revision led to Ghana becoming categorised as a lower middle income country rather than a low income country (Guardian 2012; World Bank 2015b), with a per capita GNI of $3,900 (Atlas Method PPP) in 2015 which was only just below the $4,125 level which marked the lower boundary of upper middle income countries at the time using the Atlas methodology. A point which was perhaps under-emphasised by critics of African economic statistics is that systematic and internationally comparable national income statistics have only been available for most African countries since the 1960s, and that economic statisticians are constantly seeking to improve the range and quality of statistics which are available. The reaction of Zambian statisticians is probably representative in their criticism of Jerven (Zambia 2013). This broader view can also be placed in the context of Morgenstern’s illuminating discussion of the limitations of economic statistics internationally (Morgenstern 1963) and the fact that accurate estimates of UK national income statistics date from the late-1930s (Tribe 2015, Chap. 2). The United Nations System of National Accounts, on which most national income statistics are based, was first published in 1953 (UN 1953) and the most recent edition relates to 2008 (UN 2009).

A 2011 IMF report outlined the impact of the ‘re-basing’ of the Ghanaian national accounts statistics:

Rebased national accounts for 2006–2009 were adopted in November 2010. The revisions included improved data sources, a change in the base year from 1993 to 2006, and updated compilation methodologies (the 1993 system of national accounts, and the latest version of the industrial classification standards, ISIC version 4). Under the new data, national incomes have been revised up by an average of 65 per cent during 2006–2009, while the average real growth rate for these years is about one percentage point higher than earlier estimated. On the rebased data, the gross national income per capita is estimated to have reached US$1,240 in 2010 (using the Atlas method), making Ghana a lower middle income country.

The updated accounts reflect better the sectoral composition of economic activity, and capture more effectively service sector and informal activities. The revised data highlight the increasing importance of the service sector, which accounts for about one-half of the economy and explains in part the faster growth rate in the sample period (particularly in the information/communication and financial services). (IMF 2011—Ghana IMF Country Report No. 11/128 page 86)

Another IMF report had the following to say about the state of the Ghanaian national income accounts four years after the release of the new series of data (IMF 2014a):

National Accounts: Ghana compiles annual and quarterly estimates of GDP by economic activities at current and constant (2006) prices following the System of National Accounts 1993 regularly. Incorporation of latest/updated data sources, changes in conceptual treatment/methodologies and rebasing of national accounts to 2006 led to a significant upward revision of the estimates of GDP published in November 2010. In addition, Ghana Statistical Service recently published annual GDP by expenditure estimates at current and constant (2006) prices for the years 2006–2012. Private consumption expenditure, which was estimated as a residual in the old series, has been calculated directly in the rebased series.

Ghana is one of the countries participating in the Quarterly National Accounts Module of the Enhanced Data Dissemination Initiative (EDDI) supported by the United Kingdom Department for International Development (DFID). Four missions have been conducted (September 2010, April 2011, November/December 2011, and February 2013) under this project. The February 2013 mission worked towards further improving the methodology of the current and constant price QGDP estimates by economic activity. In addition, the methodology for each component of the published annual expenditure estimates was scrutinized and a preliminary methodology for expanding these to quarterly was developed. The constant price methodology has been improved by introducing forestry stock changes and extending the range of indicators for measuring cocoa production, hotels, and air transportation. It is expected that estimates of QGDP by expenditure will be compiled by mid-2014. (IMF 2014a—Ghana: Staff Report for the 2014 Article IV Consultation—Informational Annex—page 10)

The World Bank Country Partnership agreement for 2013–2016 had the following to say about the Ghanaian statistical situation:

In reviewing its survey calendar, GSS (the Ghana Statistical Service) should consider all areas of statistical data collection and look at ways of strengthening collection of data in support of economic statistics, agriculture and other weaker or neglected areas. In economic statistics this would include support to improving data availability and coverage for national accounts and the GSS ability to re-base key economic statistics every five years. In financial statistics it would be encouraging Government to improve their availability, timeliness and comprehensiveness and commit to making these publicly available through, for example, the Ghana Open Data Initiative. (World Bank 2013c, Annex 6—Statistical Assessment, p. 148)

There are two particular areas involving comparability where clarification is necessary following the significant upwards adjustment of Ghana’s national income statistics. The first concerns time series data. Logically all national income data for years prior to 2006 would need to be updated to make them consistent with the 1993 UN SNA. This would involve a considerable volume of work entailing upgrading from the 1968 SNA to the 1993 SNA methodology, assuming that the basic data required for such an exercise were available. Such an upgrade would involve differential impacts on individual sectors and sub-sectors of the economy meaning that no easy conversion ratio could be used for such an adjustment. Figure A.3.1 shows that the proportional differences between the GDP growth rates for 2007 to 2014 as measured using the ‘old’ and ‘new’ methodologies vary significantly from year to year reflecting, inter alia, the different sectoral growth rates year by year over this period.

Fig. A.3.1

GDP growth rates for ‘old’ and ‘new’ data series, 2007–2014. Source: Republic of Ghana (2014), p. 8.

The GSS has been making good progress with the detailed revision of national accounts data. The release of revised data as it becomes available means that for some years to come there will be apparent inconsistencies between sets of national income data when comparing successive statistical publications. This applies to both Ghanaian and international publications of national income data. For example, in April 2017 a considerable amount of new GDP data was published by the GSS (2017a, b) at a time when the manuscript for this book was almost ready in a final version. It was, unfortunately, not possible to incorporate this new data into the text and tables, but the main arguments and conclusions of the book will not be significantly affected by this limitation. The release of these publications is a very encouraging sign for those following the development of the Ghanaian economy.

Jerven has been very critical of economists’ interpretation of sub-Saharan African national income statistics (Jerven 2015a, Chap. 4). It is unclear whether his interpretation is itself correct, particularly where there has been a significant revision leading to a large increase in the official data for GDP. Two quotations will illustrate this problem. First: “Andy McKay (2013, p. 51) summarizes the pattern this way: ‘From the early to mid-1990s on there is a significant reversal: aggregate per capita GDP rose by 31 per cent between 1994 and 2010, an average of 1.7 per cent a year.’ An increase in the aggregate GDP of 31 per cent is not that much if we keep these other revisions in mind. Both Ghana and Nigeria saw larger increases than 31 per cent just due to a revision in the statistical material” (Jerven 2015a, p. 112). Second: “It is likely that very recent growth data are overestimating economic growth. First, for some economies—and Ghana is the best example—the growth figures are higher because there was a recent large upward revision in GDP levels” (Jerven 2015a, p. 118).

Figure A.3.2 should make this problem clear. The line ABC represents GDP levels (and the slope of the line represents GDP growth) for the old data series. The line DEF represents GDP levels (and the growth) with the new data series. The line BE represents the increase in the level of recorded GDP arising from the revision of the GDP data (the lines ABC and DEF are parallel, and the growth rates are essentially the same). The slope of the line AF shows what the growth rate would appear to be if the estimation of the growth rate started with the old series values at A and ended with the new series values at F. Quite obviously the (erroneous) rate of growth shown by AF is higher than the rate of growth shown on ABC and DEF. It is this line AF which Jerven appears to be referring to in the above quotations—and particularly in the second quotation. The GDP growth of 31 per cent to which McKay is referring is on the ABC or DEF growth paths, while Jerven has apparently interpreted the 31 per cent growth to refer to the AF growth path.

Fig. A.3.2

Illustrative interpretation of revised GDP series. Source: Authors’ illustration.

The second clarification is that relating to international comparability of national income statistics, between Ghana and other sub-Saharan African countries, and between Ghana and the rest of the world. There is a considerable recent literature, following on from Jerven (2012), outlining the revision of national income statistics in a range of African countries (e.g., Jerven 2013a, 2014a, 2015a). The Ghanaian experience of an increase in the level of GDP by about 60 per cent has not been repeated in other countries, where increases following the adoption of the 1993 SNA have occurred. This implies, of course, that Ghana’s per capita income (and other international comparative measures of economic development and welfare) will now be higher relative to these other countries. Figure A.3.3 shows the effect of the revised statistics on comparisons of GDP per capita for a range of sub-Saharan African countries, so that for 2009 Ghana moves to a significantly higher level of per capita income than Zambia, Senegal and Kenya, while with the old data series Ghana’s per capita income was substantially lower than that for these countries.

Fig. A.3.3

GDP per capita: SSA comparisons with ‘old’ and ‘New’ Data Series for Ghana.Source: IMF (2011, p. 7, Figure 1).

Despite the discontinuity which now occurs between the GDP series up to 2006 and the series which follows that year, it is still possible to obtain a view of the relative levels of economic activity within the two series. The fact that there is an overlap for the years 2006–2010, with data from both the old and new series available for those years, makes it possible to attempt comparisons. This means that, for example, quantitative analysis of economic growth and of the relationship between variables such as the level of savings and investment, inflows of foreign direct investment, and receipts of official development assistance and the rate of economic growth can be undertaken if a few somewhat ‘heroic’ assumptions are made. However, such quantitative analysis must be undertaken with care taking cognisance of the limitations of the statistical data which are available.

Another issue associated with the interpretation of GDP (and other) data is the treatment of Ghana cedi values, particularly given the high rates of inflation which have occurred in many years during the period covered by this volume. In 2007 a ‘new cedi’ (GH₵) was adopted as the official unit of currency, being equivalent to the ‘old cedi’ (₵) divided by 10,000 (BoG 2007b, pp. 33–36) (an informative document about the history of Ghana’s currency is also available—BoG n.d.-a). Wikipedia is also an interesting source for the history of the cedi currency (Wikipedia n.d.), and Swanepoel (2015) represents another useful source. Opare-Henaku et al. (2013) report an illuminating piece of research on the response of the Ghanaian population to the introduction of the new currency. In addition to the significance of finding appropriate deflators for data series expressed in current price values, the changing international exchange values of the cedi are also an important factor in the analysis of the development of Ghana’s economy. The tables in this volume, including the Statistical Annex, have endeavoured to present some reliable and consistent data.

The treatment of ‘non-monetary’ output in the national income accounts is also significant. This was discussed in a review of economic aspects of Ghanaian and East African rural development in the late-1990s at a conference at the University of Science and Technology, Kumasi (Tribe 1998). It has not been possible to update this discussion in the context of the revision of Ghanaian national income accounts. The essence is that the level of rural output, and particularly investment, is often understated in the context of estimating both unrecorded monetary value added and non-monetary (often described as ‘subsistence’) output. The 1997/1998 review found that while East African national income accounts include separate but explicit entries for non-monetary national income (each of Kenya, Tanzania and Kenya having different ways of recording this element of national income) the Ghanaian national accounts at that time did not have separate entries for non-monetary national income but included it in the full set of accounts, not distinguishing between monetary and non-monetary elements (GSS 1996).

Judgement of the quality of Ghanaian socio-economic statistics is very relevant to the robustness of the content of this book. The World Bank’s World Development Indicators (WDI) has included two measures of the quality of statistics in recent years. The first of these is the “methodology assessment of statistical capacity” and for 2004 to 2006 Ghana was given a value of 30 on a scale from 0 to 100, for 2008 and 2009 the value was 40, for 2010 to 2012 the value was 50 and for 2013 to 2016 the value was 60 (World Bank 2015a, 2016a). The conclusion from this data must be that from a comparatively poor position in the mid-2000s Ghana’s statistical capacity is judged to have improved significantly, but by 2016 still had ‘room for improvement’. The second measure is for the “level of statistical capacity” (which is not available from the 2016 edition of the WDI) and for Ghana this improves steadily from about 51 (again on a scale from 0 to 100), reaches a peak of 65.6 in 2010, and then declines slightly to just over 62 in 2015 (World Bank 2015a, 2016a). Both sets of data lead to similar conclusions, but it would clearly be necessary to explore the definitions behind these measures and the methods of derivation before drawing any more robust conclusions.