New Political Economy

ISSN: (Print) (Online) Journal homepage: https://www.tandfonline.com/loi/cnpe20

Imputing Away the Ladder: Implications of

Changes in GDP Measurement for Convergence
Debates and the Political Economy of
Development

Jacob Assa & Ingrid Harvold Kvangraven

To cite this article: Jacob Assa & Ingrid Harvold Kvangraven (2021): Imputing Away the Ladder:
Implications of Changes in GDP Measurement for Convergence Debates and the Political Economy
of Development, New Political Economy, DOI: 10.1080/13563467.2020.1865899

To link to this article: https://doi.org/10.1080/13563467.2020.1865899

Published online: 07 Jan 2021.

Submit your article to this journal

Article views: 419

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at

https://www.tandfonline.com/action/journalInformation?journalCode=cnpe20
NEW POLITICAL ECONOMY
https://doi.org/10.1080/13563467.2020.1865899

Imputing Away the Ladder: Implications of Changes in GDP

Measurement for Convergence Debates and the Political
Economy of Development
Jacob Assaa and Ingrid Harvold Kvangravenb
a
The New School, New York, USA; bUniversity of York, York, UK

ABSTRACT KEYWORDS
What are the implications of changes in measurement standards of GDP Economic statistics; GDP;
for global convergence debates? What are the political economy convergence; global growth;
implications? To answer the former question, we examine the changes Rise of the South
in national accounting standards from the early 1990s. Revisions to the
System of National Accounts (SNA) – the international standard for
constructing GDP – include several major changes to how production is
measured, including the reclassification of financial intermediation
services, R&D, and weapons systems as productive activities – all areas
in which countries in the West has had an advantage in recent decades.
In addition, there has been an increase in the proportion of imputations
in the 1993 and 2008 revisions, which privileges the economic
structures of the West. Overall, we find that these changes have had the
effect of boosting the GDP of the West relative to the rest of the world
and thus to an underestimation of global convergence compared to
previous measures of GDP. To answer the second question, the paper
unpacks the political economy implications of national accounting
standards favouring Western economies along several axes, including
the impacts on voting shares in international institutions, domestic
policy incentives and epistemological debates about sustainable
development.

Introduction
On the one hand, despite the important emerging literature critically assessing the political economy
of GDP measurement, the implications for convergence debates of a GDP defined according to
Western economies’ structures has not yet been scrutinised. On the other hand, while measurement
of convergence is a rich field of research, that debate has not paid adequate attention to how
changes in GDP measures impact perceived growth rates. This article addresses this gap in the
two literatures by assessing the political economy implications of GDP measurements for develop-
ment and by analysing how the choice of output measure affects the estimated rate of global con-
vergence. To do so, we examine the changes in national accounting standards from the early 1990s
in the wider context of global economic restructuring. Starting in the 1980s and accelerating in the
1990s and 2000s, developed countries have experienced a structural transformation, with many
manufacturing activities, and some services, outsourced to developing countries, and an increasing
share of value-added concentrated in high-end service sectors such as banking, real-estate, insur-
ance, intellectual property, research and development, as well as production of military weapons
systems. Meanwhile, the key measure of growth – GDP – has been redefined in ways that favour

CONTACT Ingrid Harvold Kvangraven [email protected]

© 2021 Informa UK Limited, trading as Taylor & Francis Group
2 J. ASSA AND I. H. KVANGRAVEN

these new areas of specialisation of developed economies, while developing countries are taking
over the manufacturing mantle. More concretely, in 1993 and then again in 2008, the standards
for measuring GDP (and by derivation, growth) have been changed in a direction which gives
more weight to sectors dominant in countries in the West. While it may not be surprising that the
System of National Accounts (SNA) framework – which determines how GDP is recorded – responds
to changing economic structures in advanced economies,1 the implication of this for debates on
global economic convergence has not yet been explored. We find that the changes have had a signifi-
cant impact on the assessment of comparative growth among countries in the West and the Rest –
kicking away the statistical ladder, so to speak.2 Had the changes in GDP not been made, the devel-
oping world would appear to have converged on the developed world to a larger extent. This
paper does not engage in the ultimately normative discussion about how GDP should be measured,
but rather lays out the implications of recent changes on the assessment of convergence.
To date most critiques of GDP focus on countries which have achieved a certain level of high
income and seek to broaden the scope of GDP from a narrow economic view to a wider, well-
being related framework, by questioning the measure for what it does not account for, such as
unpaid labour, environmental degradation, happiness, freedom, welfare, rather than what it does
measure. In fact, in the development literature, GDP is often considered at least to be ‘constrained
by a consistent economic theory’, in contrast to some of the measures that try to incorporate social
and environmental goals into a broader measure of social and economic wellbeing (Felice 2015, p. 2).
However, economic theory has had only a weak influence on the development of national account-
ing in general and GDP in particular (Bos 1995, Mitra-Khan 2011, Assa 2018), despite the fact that
GDP remains the most used and most respected accounting measure (Seers 1972, van der Bergh
and Verbruggen 1999, Felice 2015).
Changes to national accounting standards also have implications for theoretical and empirical
debates about the source of such growth as well as serious real-world political economy impli-
cations. For example, the appearance of success in fast-growing countries can inspire other countries
to imitate their policies. Depending on the measure of output used, different countries will appear to
grow faster, given their particular economic structure. Furthermore, a country’s economic size rela-
tive to world output contributes to the determination of its voting rights in international organis-
ations such as the World Bank and IMF, and its level of per capita GDP determines its eligibility
for concessional foreign aid. Finally, output in both level and growth enters certain sensitive
ratios such as debt-to-GDP and deficit-to-GDP, which are central to political debates as well as inter-
national treaties (such as the Maastricht Treaty establishing the Euro currency).
Section 2 discusses the political economy of GDP measurement generally, and the convergence
debates specifically – debates where GDP is employed uncritically as a measure of growth. Section 3
documents the changes to the SNA in 1993 and 2008 and demonstrates how they have contributed
to moving the developmental goal-post in favour of developed countries’ newly-found areas of
specialisation, largely by adding more imputations to the measure of output. In Section 4 we
build on Basu and Foley’s (2013) Narrow Measured Value Added (NMVA), which uses the criterion
of whether net output in a certain sector is directly measurable or instead imputed from net
income (NMVA includes only the former).3 This allows us to analyse groups of countries – both by
region and by income level – and revisit some of the key convergence debates in light of the shifting
goal-post of GDP. This is a particularly timely debate, as future revisions of GDP are likely to include
more imputations in the form of ‘intangibles’, thus intensifying the trend noted in this article
(Corrado et al. 2017, Fraumeni et al. 2017, Hulten and Nakamura 2020). Section 5 examines the pol-
itical economy implications of including imputations in GDP and demonstrates that the relative
growth of countries’ NMVA paints a different picture of relative growth. This change in measure
changes the convergence assessments of countries and regions, and also the timeline for some
countries’ projected catch-up with the economic frontier. Both these factors play a key role in inform-
ing national and international policies, as the way growth is defined affects what is considered econ-
omic success. Section 6 concludes.
 NEW POLITICAL ECONOMY 3

The Political Economy of GDP Convergence

While political economy scholarship has in recent years provided much-needed critical analyses of
GDP as an objective indicator of progress (Irvine et al. 1979, Coyle 2014, Lepenies 2016, Masood
2016), the impacts of changes in GDP measurement on the convergence debates and the perception
of progress in the non-Western world has largely gone unstudied (as also recognised by Mügge
2016). While political economists will highlight the social construction of economic indicators and
the way they change over time, economists tend to point to reform in indicators as necessary ‘mod-
ernisation’ to keep the indicators in line with contemporary structures of the economy (e.g. Lequiller
and Blades 2006, p. 3). From the latter perspective, GDP has developed to reflect structural economic
and political transformations, such as the rise of economic planning in the 1930s (Perlman and Mar-
ietta 2005) and contemporary financialization in recent times (Christophers 2011, Assa 2017).
However, even to the extent that GDP has evolved with changing economic structures, it is
always the economic structures that are prevalent in the West that are considered the appropriate
yardstick. Despite a massive structural gap between economies at the ‘frontier’ and low-income
countries, for example, all countries are expected to more or less comply with international stan-
dards of measurement of national accounts. This has been the case since colonial powers and gov-
ernments of newly independent states institutionalised macroeconomic measurement systems of
the West, which already then were often not well suited to the local economies (Jerven 2012b),
leading to unreliable data (Jerven 2013, Woods 2014).
This article intervenes in these debates both to address what the implications of this Western-
oriented GDP measure are both for how we perceive growth and convergence and for how the chan-
ging perceptions of convergence and growth, in turn, have serious political economy consequences.
The implications stretch across a variety of axes, including domestic policy-making, voting rights in
international institutions, and the convergence debates more generally.
Turning to the convergence debates first – the one thing that all sides in the convergence debates
have in common is their use of the growth of GDP per capita as the main yardstick. The traditional
debate about convergence before the 1980s was largely about the ‘Great Divergence’, which took
place roughly between 1500 and 1950, between ‘the West’ (Western Europe and its former settler
colonies in North America and Australasia) and ‘the Rest’, in terms of income per capita (Popov
and Jomo 2018). According to the Maddison Project (2013), the ratio between the West and the
Rest rose from 1:1 in 1500 to 6:1 in 1900. The disparity was roughly stable between the 1950s
and 1970s, but increased, on average, during the Washington Consensus era, before falling slightly,
but in an uneven manner. The ratio remained at around 6:1 throughout the 1900s.4
By the 1980s, economists started empirically exploring the question of convergence and linking
the findings to different growth theories (see for example Kormendi and Meguire 1985, Baumol
1986, DeLong 1988, Grier and Tullock 1989, Barro 1991, Barro and Sala-i-Martin 1992, Mankiw
et al. 1992, Holtz-Eakin 1993). The neoclassical Solow (1956) and Swan (1956) models predict ‘absol-
ute convergence’ – that poor countries with small capital stocks will grow faster than rich countries
with large stocks. Here, absolute convergence is driven by diminishing returns to capital implicit in
the neoclassical production function, which leads to the prediction that the rate of return to capital is
large when the stock of capital is small and vice versa. Endogenous growth models contradicted this
absolute convergence thesis, as they rely on the existence of externalities, human capital, and
increasing returns, opening up the possibility of no convergence, or even divergence (see e.g.
Romer 1986, Lucas 1990, Rebelo 1991). Therefore, it was widely believed in the 1990s that testing
for convergence was also an exercise in testing competing growth theories. As these debates
evolved, there were concerns with possible misinterpretation of available data and especially
measurement errors when it comes to measures of productivity (e.g. Griliches 1994), but little atten-
tion has been paid to how changes in GDP measures impact the convergence debates.
While Pritchett (1997, p. 3) wrote that divergence in living standards is ‘the dominant feature of
modern economic history’, there have been studies since then that suggest that some degree of
4 J. ASSA AND I. H. KVANGRAVEN

convergence between poor and rich countries has taken place since the end of the 1990s especially
among ‘clubs’ of countries (Ocampo et al. 2007, Lin and Rosenblatt 2012, Mazumdar 2016, Popov and
Jomo 2018). While some suggest that the Great Divergence is potentially about to end or even
reverse (e.g. Nayyar 2013), the main consensus is still that there is no strong trend of gradual
global convergence in income levels (Islam 2003, Popov and Jomo 2018, Johnson and Papageorgiou
2020).
What are the real political economy implications of this ‘blind spot’ in the debates on conver-
gence? It has long been recognised that there are geopolitical and political economy implications
related to the measurement of GDP (Coyle 2014). One reason a country may want to inflate its
GDP is to show economic strength. For example, Petty and Gregory King in England, and Pierre
de Boisguilbert and Marshall Vauban in France, all used their estimates of national income to demon-
strate their countries’ fiscal strength, and their estimates were accompanied by calls for certain
growth policies (Assa 2015). Furthermore, GDP became increasingly important in the twentieth
century, first to measure the recovery from the Great Depression, and then as a measure to show
how the US was out performing the Soviet Union during the Cold War (Philipsen 2015). Today,
the heated debates about the actual size of China’s GDP confirms that relative economic strength
is still of utmost importance (Toma et al. 2017). It is still common practice among policymakers
and politicians to use economic indicators to assess policies – and often in comparison to other
countries (Fougner 2008, Davis et al. 2012, Hansen and Porter 2012, Mügge 2016, 2020). Even
when there are no formal implications, country-state rankings can induce domestic reforms
(Cooley and Snyder 2015, Kelley and Simmons 2015).
However, there could also be economic incentives for low-income countries to remain ‘low
income’ in order to access concessional lending from the World Bank, the IMF and bilateral
lenders or to pay lower contributions to international organisations (consider also China’s late adop-
tion of the World Bank’s International Comparison Program). While Jerven (2012a) finds that low-
income and lower-middle income countries often make it a national priority to move up the
income-ladder, which in turn is thought to attract funds from commercial banks and investors,
the extent to which this is true for a given country at a given time will depend on the political
and economic context. Furthermore, political leaders may have incentives to show to their populace
that they can foster economic development, which a transition to a higher income-category could
do. For example, before Nigeria revised its GDP in 2010, it had announced that middle-income status
was a ‘political target’ (Jerven 2012a, p. 140).
Given the rebalancing of the global economy towards ‘rising powers’, their politics of measure-
ment may play a larger role in future harmonisation efforts.5 The more specific implications of the
recent changes in GDP measurements on political economy outcomes such as voting shares in inter-
national institutions, measurement of the Sustainable Development Goals, and membership of the
European Union (EU) are considered below, after having laid out what the recent measurement
reforms entail.

Deregulating GDP
While there have long been critics of GDP as a measure of wider dimensions of human well-being
such as happiness, unpaid care work (Waring 1990, Folbre 2015), environmental costs (Hickel
2019), human development (Stiglitz et al. 2010) and quality of life more generally (Fioramonti
2013, Fleurbaey and Blanchet 2013, Tavernier et al. 2015, Decancq and Schokkaert 2016, Jones
and Klenow 2016, Wesselink et al. 2007), this literature has not paid much attention to the significant
changes in the economic methodology of constructing GDP over the past 25 years. Meanwhile, the
studies of changes in GDP methodology that do exist focus on the general implications, or the
specific implications for advanced economies (e.g. Christophers 2011, Assa 2017, Feldstein 2017),
rather than the implications of GDP methodology for the political economy of development and
the convergence debates.
 NEW POLITICAL ECONOMY 5

In 1993 and then again in 2008, there were reforms to the SNA that led to significant changes of
the location of the so-called production boundary, which determines what is included in GDP and
what is not. Many economic activities – financial intermediation, research and development and
the production of weapons – were previously excluded from GDP as either non-productive or as con-
stituting productive inputs to other outputs (hence deducted as intermediate consumption). The
inclusion of these economic sectors in the production boundary since 1993 and 2008 has added dis-
proportionately to the GDP of developed countries, which have in recent decades specialised in
these activities and moved away from traditional pillars of development such as manufacturing
and infrastructure-related services.
A full history of national accounting is beyond the scope of this paper, and has been treated else-
where (Studenski 1958, Bos 1995, Vanoli 2005, Assa 2015). However, it is important to note that the
measurement of the income and growth of nations has been historically and geopolitically contin-
gent from its seventeenth century beginnings, with critical changes in the structure and substance of
the measurements at each step along the way (Assa 2015, 2017). As Seers (1972) observed more than
four decades ago: GDP is not a value-neutral measure at all. From the 1920s onwards, national gov-
ernments took over control of the measurement of output, an endeavour that was previously carried
out by individuals (Kendrick 1970, Bos 1995, Vanoli 2005). Along with this change in the institutional
setup of measurement came a new focus on how much output nations could produce, rather than
just on their income and expenditure (Vanoli 2005). Keynes played an important role in pushing for
the publication of the national accounts in the form of Gross National Product, based on his demand
for an output aggregate, which he argued needed to include government expenditures next to
private consumption and investment, as well as add up to the total sum of taxable income
(Keynes 1940, Tily 2009, Mitra-Khan 2011).
The concept of production is defined in the most recent of the three approaches to calculating
GDP, namely value-added. This concept – unlike the income or expenditures approaches to
GDP – requires a production boundary to determine what is to be included in GDP. As the new dis-
cipline of national accounting became an international standard of the United Nations (UN) in 1953,
the methodology was set by a group of chief government statisticians under the aegis of the UN
Statistical Commission. At the end of the Cold War there was a second shift in the institutional
responsibility for the System of National Accounts. While the UN had been the sole custodian of
this standard up to that point, in 1993 the UN was joined by the World Bank, the International Mon-
etary Fund (IMF), the Organization of Economic Co-operation and Development (OECD), and the
European Union (EU) under the aegis of an inter-secretariat working group on national accounts
(ISWGNA). In the first two institutions, voting shares are generally determined by proportion of
financial contributions as well as economic weight, while the last two represent developed countries.
This geopolitical shift also reflects a departure from the traditional role of national governments
(directly or through the UN) as custodians of national statistics, towards a greater role for financial
institutions, both national and international.
Following the institutional shift in power was a substantive change in the structure and definition
of GDP, consisting of widely expanding the production boundary to include many activities that
were hitherto considered either external to production or at most as intermediate inputs to it. His-
torically, these shifts in the production boundary have not been based on progress in economic
theory but rather developed through a ‘techno-political process’ (Bos 1995, Christophers 2011,
Assa 2017, 2018). From a developing country perspective, the standards that have emanated from
the SNA have been perceived as largely top-down (Ward 2004, Fioramonti 2013).
GDP has become more imputation-heavy with the revisions of the past decades. Imputing value-
added from incomes differs in important ways from measuring it directly. In the latter case, both
gross output and intermediate inputs are deflated using a price index (to account for inflation),
and real value-added equals real gross output less real intermediate inputs. The deflation ensures
that while the price of a car increases, for example, real value-added accounts for only the quantity
of cars produced. When value-added is imputed based on net incomes (profits and wages), by
6 J. ASSA AND I. H. KVANGRAVEN

contrast, the formula degenerates to deflating gross revenues and deflating costs, then deducting
the latter from the former. There is no accounting for a quantity here, despite adjusting for
inflation, since there is no direct measure of ‘output’ to begin with. Thus, quantity and price are inse-
parable for imputed services, since net income is a monetary concept by definition.
The rest of the section describes specific areas of change in the SNA revisions in 1993 and 2008,
both on the consumption side (exemplified by financial intermediation) and the investment side
(exemplified by intellectual property and military spending). As mentioned above, these changes
have had differential impacts on developed and developing countries, as we will demonstrate
with examples throughout the section.

Imputations Affecting Consumption

In this section we consider how changes to how financial intermediation is measured affect the
recorded value of consumption. The 1968 SNA treated the intermediation activities of banks – for
which income is derived from interest differentials rather than explicit fees – as the input of a
notional, or ‘imaginary’, sector, and thus it was deducted from total value-added and did not
affect GDP. While this statistical fiction of assigning banking output as an input to an imaginary
sector was bizarre, the SNA 1993 went further and allocated the part of this income stream paid
by households to final consumption (while treating the part paid by firms as intermediate consump-
tion, again deducted from total value-added). In a stroke of a pen, this methodological change made
finance (to households) productive (Christophers 2011).
In the process referred to as ‘financialisation’, often accompanied by deindustrialisation, many
developed countries have moved away from manufacturing and towards financial and real-estate
activities. Thus, the share of this imputation (known technically as Financial Intermediation Services
Indirectly Measured or FISIM) has grown much faster in these countries than in developing countries.
While in most OECD countries financial intermediation (paid for by households) adds between 1 and
2 percentage points to GDP, in the US this was 3.4 per cent in 2011.
The SNA 2008 did away with the requirement that only interest on intermediated funds be
included in GDP and required the inclusion of all loans and deposits used to impute FISIM (UN
2009, 583:A3.25). In other words, even banks’ own money could now be used to create such ‘pro-
duction’, without the pretext of providing an intermediation service. As 80 per cent of bank loans
in the US go to finance the purchase of real-estate or financial assets rather than productive activities,
the SNA 2008 also made speculation productive.
Since the 1970s, the importance of interest income for banks has declined, as much more of their
revenues now come from fees (anything from overdraft charges, late fees, mortgage origination fees,
to underwriting and foreign exchange fees). Although the fee income of banks is explicit (rather than
implicitly estimated using the FISIM formula), the value-added of the financial sector based on these
fees is still an imputation, based on deducting fees paid by banks from fees received by them.
Together with similar charges in the real-estate and insurance industries, these so-called FIRE
sectors (Finance, Insurance, and Real Estate) account for nearly a quarter of GDP in OECD economies
in the twenty-first century (Figure 1).
The financialisation of the US economy and of other developed economies thus has far-reaching
implications for the assessment of their growth rates, especially in comparison to developing
countries where finance has not exploded in a similar manner. Figure 2 shows the weight of the
FIRE sector in the US and Chinese economies, and Figure 3 shows the impact this differential has
on their convergence when FIRE is included or excluded.
It is true that, in either case, China absolute per capita income places it in the middle-income
group while the US is in the high-income group. However, as development and convergence also
have a relative dimension, the role of imputation in assessing performance is important. The
growth rate of China’s per capita GDP without the FIRE sector is 3 times that of the US, compared
to only 2.5 times including FIRE. And if we excluded all imputations (using NMVA), the ratios are
 NEW POLITICAL ECONOMY 7

Figure 1. The Finance, Insurance and Real-Estate (FIRE) sectors as per cent of GDP in the US vs. China (constant prices). Source:
United Nations, Main Aggregates and Detailed Tables (MADT) database, Table 202, http://data.un.org/Data.aspx?d=SNA&f=
group_code%3a202 (last accessed March 2020).

9.2 vs. 6.3, respectively. This is a difference between China doubling its per capita income every 7.6
years or every 11 years, a key correlate of economic convergence.

Imputations Affecting Investment

Both measures of military expenditure and intellectual property have been changed substantially
over the past reforms of SNA. The SNA 68 did not include military expenditures on fixed assets as
investment, except those related to constructing or modifying family dwellings for military person-
nel. By contrast, SNA 93 included all military expenditures ‘which could be acquired by civilian users
for purposes of production and that the military uses in the same way’ (UN 1993, 660:70). Thus docks,
roads, airfields etc. were included, while weapons or vehicles and equipment solely used to launch
them were still excluded. The idea was that the former were potentially also productive assets while
the latter were only destructive in nature.
The 2008 SNA knocked down the boundary further, by doing away with this distinction, and
included expenditures on weapons systems in government investment in fixed assets. This
change inflated the GDP of weapons-producing countries. Looking at military expenditure in con-
stant prices reveals how both high and middle-income countries have pulled far away from low
income countries. The inclusion of weapons systems in GDP is thus inherently biased against the
poorest countries (which may be large consumers of weapons, large and small, but produce very
little of them), as shown in Figures A1–A3 in Annex III. To complicate things further, measurement
practices may differ across countries (Mügge 2016). For example, while the US has historically
included military expenditure in its GDP calculation, most other countries have not, thereby
leading to an overstatement of US GDP (Lequiller and Blades 2006).
Furthermore, developed countries have a lead in activities resulting in intellectual property pro-
ducts (previously named ‘intangible produced assets’), including R&D, computer software and
8 J. ASSA AND I. H. KVANGRAVEN

Figure 2. Impact of FIRE imputation on per capita GDP convergence between China and the US. Source: United Nations, Main
Aggregates and Detailed Tables (MADT) database, Table 201, http://data.un.org/Data.aspx?d=SNA&f=group_code%3a201 (last
accessed March 2020).

databases, entertainment, and literary and artistic originals. Before 2008, research and development
activities – which are intended to improve efficiency and productivity and thus have the potential to
increase future output – were treated as part of intermediate consumption, thus not included in
GDP. The SNA 2008 changed this classification based on the idea that R&D increases the stock of
knowledge and enables it to create new applications. This change has had a much bigger impact
on the GDP of countries creating new technologies rather than those adapting them (Figures A4–
A6 in Annex III).
While there is a wide variation among developed countries in terms of their spending on R&D as a
percentage of GDP, the relationship is clearer among developing countries. Countries with average
per capita GDP (2011–2015) of less than $11,500 spend just 0.3 per cent of GDP on R&D. Below
$23,000, only China ($11,962) and Brazil ($15,112) spend more than 1 per cent (2 per cent and 1.2
per cent respectively). At income levels over $40,000, on the other hand, only a handful of
countries spend small percentages, mostly oil-rich countries or tax havens (Oman, Bahrain,
Saudi Arabia, Bermuda, UAE, Kuwait and Qatar). The rest of the $40,000 plus group average 2.3
per cent R&D in GDP.

Redefining Growth
Each of the above revisions to the national accounting methodology has added disproportionally to
the perceived output (as measured by GDP) of more developed countries. To see the combined
effect of these methodological revisions on estimated growth rates, we now look at a selection of
countries for which data is available in contemporaneous GDP series.6 The top panel of Table 1
shows that shifting to the new SNA had differential impacts even for the growth rates of developed
countries (the years selected were the ones where data for all three SNAs was available for these
 NEW POLITICAL ECONOMY 9

Figure 3. Breakdown by Expenditure of Changes in SNA on Bangladesh’s 1992 as per cent of Nominal GDP. Source: United
Nations, Main Aggregates and Detailed Tables (MADT) database, Table 101, http://data.un.org/Data.aspx?d=SNA&f=group_
code%3a101 (last accessed March 2020).

countries). While these differences in average annual growth rates may seem small, their cumulative
impact over 27 years is significant.7
The differences are even more dramatic among developing countries and looking at cumulative
growth rates over several years for which these countries all have data in both SNAs (bottom panel of
Table 1). SNA 1993 not only benefited Korea but penalised Bangladesh and Bhutan. In the case of
Bangladesh, the moderate increases in its 1992 household expenditure and business investment
were outweighed by the 60 per cent decrease in government expenditure.
Looking at Korea’s convergence with the US in more detail is illustrative. Both countries have par-
allel data series for SNA 68 and SNA 93 from 1970 to 1997, allowing us to view the drastic impact of
revising the SNA on their relative growth rates. Using as starting figures for 1970 data on per capita
income from the Maddison Project ($2,989 for Korea and $23,958 for the US), Korea’s per capita

Table 1. Average annual and cumulative GDP growth rates for three SNA systems (ISIC 3.1), selected countries.
Annual growth rate Cumulative growth
SNA 1968 SNA 1993 SNA 2008 SNA 1968 SNA 1993 SNA 2008
1970–1996
Australia 3.1% 3.2% 126% 136%
Korea 8.0% 8.9% 701% 889%
Mexico 3.6% 3.6% 3.6% 157% 157% 157%
Turkey 4.3% 4.3% 210% 210%
USA 2.5% 2.9% 2.9% 98% 122% 124%
UK 2.0% 2.1% 72% 77%
Israel 4.6% 4.8% 224% 241%
1992–1997
Bangladesh 4.2% 3.9% 28% 26%
Bhutan 5.4% 4.5% 37% 30%
Korea 5.9% 6.1% 41% 43%
10 J. ASSA AND I. H. KVANGRAVEN

Figure 4. Korea-US Convergence under SNA 68 vs. SNA 93. Source: United Nations, Main Aggregates and Detailed Tables (MADT)
database, Table 102, http://data.un.org/Data.aspx?d=SNA&f=group_code%3a102 (last accessed March 2020).

income in 1970 was 1/8th (or 12.5 per cent) that of the US. By 1997, it has grown to just under 40 per
cent according to SNA 68, but over 55 per cent according to SNA 1993 (Figures 4 and 5).
In this case the move to SNA 93 favoured Korea relative to the US. This can be disaggregated by
the differential impact of the SNA revision on both countries, as is illustrated in Figure 6.
However, even though the initial boost to US investment by changing to SNA 93 was bigger than
the comparable increase in Korea, over time Korean R&D (which is recorded under investment)
increased much faster as a per cent of GDP than in the US (see Figure 7).

Imputed Growth: GDP vs. Measured Value-added

In the absence of consistent counter-factual data on pre-SNA-93 output for most countries, the
Narrow-Measured Value-Added (NMVA) measurement of output (as defined by Basu and Foley
2013, Foley 2013) serves as a useful proxy for an indicator of output without imputations. The deli-
neating criteria in Basu and Foley (2013) is not that of productive vs. non-productive activities (remi-
niscent of classical political economy) but rather on which goods and services it is possible to directly
measure value-added for, rather than impute based on incomes. NMVA is an indicator that only
includes industries with directly observable output, such as agriculture; mining; utilities; construc-
tion; manufacturing; wholesale trade; retail trade; transportation and warehousing; storage and com-
munications. Value-added in these sectors is measured as the monetary value (i.e. the quantity of
goods and services times their price) of the actual output produced less the monetary value of
the intermediate inputs used. For other industries, while efforts have been made to move to
output-based measures, this is not entirely possible given the intangible nature of their outcomes.
In GDP, value-added for these sectors is imputed (partly or wholly) based on their net incomes (rev-
enues less costs).8 It is important to note that NMVA is far from perfect. While not including many
 NEW POLITICAL ECONOMY 11

Figure 5. Average (1994–1996) Percentage Increase by Component, Change from SNA 68 to SNA 93. Source: United Nations,
Main Aggregates and Detailed Tables (MADT) database, Table 101, http://data.un.org/Data.aspx?d=SNA&f=group_code%
3a101 (last accessed March 2020).

Figure 6. R&D as per cent of GDP, Korea and the US. Source: Authors’ calculations based on World Bank, World Development
Indicators database, series GB.XPD.RSDV.GD.ZS. Accessed March 2020.
12 J. ASSA AND I. H. KVANGRAVEN

Figure 7. China-US Convergence (GDP and NMVA). Source: Authors’ calculations based on GDP shares from the United Nations
Main Aggregates Database (https://unstats.un.org/unsd/snaama/dnlList.asp) and GDP in $PPP from the World Bank.

imputations as GDP, it still shares some of its other weaknesses, such as not accounting for unpaid
care work or the environmental costs of material production. Furthermore, it may be even farther
away from a measure of welfare than GDP, since it does not even include expenditure on services
such as health and education.
Nonetheless, the inclusion of imputations such as financial services in GDP is one reason why this
measure of output has diverged dramatically from aggregate employment in recent decades (Basu
and Foley 2013, Assa 2017). This has manifested itself in spurious mysteries such as ‘jobless recov-
eries’, on the one hand, and ‘severe job-loss downturns’ – in which more jobs are lost than standard
output-employment dynamics would predict. This distortion has created ‘productivity miracles’,
since labour productivity is defined as output (GDP) divided by (labour) input. If the numerator is
inflated, productivity may be overestimated. Annex I provides a formal analysis of productivity,
demonstrating how the current measure of productivity in the financial sector may be more of a stat-
istical artefact than a real phenomenon.
The share of imputed value-added in the US has increased far more than in China, as is shown in
Figure 8. In turn, this has had important consequences for convergence between the two countries,
depending on whether one looks at GDP (which includes the imputations) or NMVA (which does
not).
Looking at China’s convergence with the US economy with and without imputations, both the
initial gap and the rate of convergence are very different. Standard GDP shows China starting in
1990 with an economy one fifth that of the US, while in terms of NMVA it was more than a third.
Furthermore, while China is said to have taken over the US in 2014 in PPP terms, when its
economy was 17.3 trillion international dollars vs. 16.5 trillion in the US, excluding imputations con-
vergence occurred already in 2008. This is in large part due to the differential impacts of the financial
crisis of 2007–8, not just across countries, but also within developed economies. While US GDP con-
tracted −0.1 per cent and −2.5 per cent in these two years, respectively, the US economy as
 NEW POLITICAL ECONOMY 13

Figure 8. Measured and Imputed Value-Added as per cent of Total, West and the Rest, NMVA and GDP, 1990–2018. Source:
Authors’ calculations based on GDP shares from the United Nations Main Aggregates Database (https://unstats.un.org/unsd/
snaama/dnlList.asp) and GDP in $PPP from the World Bank.

measured by NMVA lost −1.2 per cent and −6.8 per cent, respectively, of measured output. Overall,
the share of industries for which value-added is imputed (government, FIRE, and other services) has
increased, but the increase was most dramatic in the West (defined as Western Europe, Australia,
Canada, New-Zealand, the US and Japan), where by 2009 more than half of GDP was thus imputed.
Thus, the reclassification of several imputed sectors from intermediate inputs to final outputs has
an inherent bias towards developed economies with a particular economic structure. The imputa-
tions’ bias towards richer countries can also be seen in the correlation between the log of per
capita GDP (in PPPs) and the percentage of GDP that is imputed (Figure 9).
Looking only at NMVA gives a very different picture of convergence in per capita income.
Figure 10 shows the differences in convergence between the West and the Rest using the two
GDP methodologies.
As Durlauf et al. (2005) discusses, growth convergence measured through econometric tests has
become a significant research industry. To speak to this, we run simple regressions of both uncondi-
tional and conditional convergence with different measures of GDP to demonstrate the significance
of recent changes. Tables 2 and 3 present the results of growth regressions for both unconditional
and conditional convergence, using GDP and NMVA, respectively. The dependent variables in both
tables are the average annual growth rates of per capita income from 1990 (the first year for which
PPP data are available) until 2017. Annex IV shows the results of convergence regressions using
market-exchange rates from 1970 to 2017. These regressions likewise show a stronger convergence
with narrow-measured value-added than with GDP. In Table 2, only the initial level of income
(output) in 1990 is taken as an independent variable, while in Table 3, additional ‘conditions’ are
included, such as the rate of population growth over the period, the average investment (gross
capital formation), log of initial population (to account for economies of scale), and the log of
initial human capital (enrolment in primary education).
14 J. ASSA AND I. H. KVANGRAVEN

Figure 9. Share of Imputations in GDP and Log GDP per Capita, 2018.

Figure 10. Per Capita Income in the Rest as per cent of the West, alternative measures. Source: Authors’ calculations based on
GDP shares from the United Nations Main Aggregates Database (https://unstats.un.org/unsd/snaama/dnlList.asp) and GDP in
$PPP from the World Bank.
 NEW POLITICAL ECONOMY 15

Table 2. Unconditional convergence 1990–2017 (market exchange rates).

GDP NMVA
Log per capita income in 1990 −0.333 −0.515
significance *** ***
R2 0.044 0.097

Table 3. Conditional convergence 1990–2017 (PPP exchange rates).

GDP NMVA
Log per capita income in 1990 −0.556*** −0.726***
Population growth −0.219* −0.118
Average Gross Capital Formation (%) 0.100*** 0.105***
Log population in 1990 0.150** 0.160**
Log of enrolment in primary education 1990 1.080** 1.043**
R2 0.270 0.307
Source: Authors’ calculations based on data from UNSD and World Bank.

In both cases, output measured by NMVA shows a stronger convergence compared to GDP. This is
of course an average of many different countries in both groups. Looking at one country such as
China has more dramatic results. NMVA per capita convergence between China and the US is 1.5
times faster than per capita GDP convergence (Figure 11).
As Figure 12 demonstrates, the convergence of NMVA not only starts from a higher level (with the
rest comprising 19 per cent of the West’s per capita income in 1990 as opposed to 14 per cent with
GDP), but also proceeds at a faster rate. The average annual growth rate of NMVA between 1990 and
2016 was 2.6 per cent for the Rest and 0.7 per cent for the West, compared to 3.1 per cent and 1.2 per
cent using GDP, respectively. That is, NMVA per capita was growing 3.7 times faster in the developing

Figure 11. China-US Convergence in per capita output, alternative measures. Source: Authors’ calculations based on GDP shares
from the United Nations Main Aggregates Database (https://unstats.un.org/unsd/snaama/dnlList.asp) and GDP in $PPP from the
World Bank.
16 J. ASSA AND I. H. KVANGRAVEN

Figure 12. Convergence between the West and the Rest, GDP and NMVA. Source: Authors’ calculations based on GDP shares
from the United Nations Main Aggregates Database (https://unstats.un.org/unsd/snaama/dnlList.asp) and GDP in $PPP from
the World Bank.

world, compared with only 2.5 times faster for GDP. At this rate, per capita GDP of the Rest would
converge with that of the West only in 77 years, as opposed to 66 years for NMVA.
NMVA per capita grew 3.8 per cent annually in the BRICs vs. 0.8 per cent in the US. In both cases
these rates are lower than GDP growth (4.5 per cent and 1.4 per cent respectively), but in relative
terms, the NMVA per capita growth of the BRICS was 4.5 faster than that of the US, compared to
only a factor of 3.3 when using GDP as a measure. Furthermore, the divergence between the two
measures first declines in the 1990s when it averages around 4 percentage points, accelerates to
5–7 p.p. from 2001 to 2008, and even accelerates even more after the crisis to 9 per cent (Figure 13).

Political Economy Implications of Recent Reforms

How do the changes outlined above matter for the political economy of development, in concrete
ways? In this section, we look specifically at how the change in GDP measurement impacts developing
economies. As Hansen and Muhlen-Schulte (2012, p. 455) argue, the use of indicators in global govern-
ance is by no means limited to knowledge sharing and learning, but amounts to a ‘mechanism of
inclusion and exclusion from complex social hierarchies’. As this section demonstrates, GDP as an indi-
cator of growth plays a similar role. At the international level, an obvious example of institutions in
which GDP measurement matter are the IMF and the World Bank. How votes in these institutions
are determined vary by agency, but economic weight is one of the factors. While Wade (2013)
argues that the Western countries keep the lead in international organisations by deliberate manipu-
lation of how votes are distributed, we identify a subtler bias. Reforming how GDP is measured to fit
the strengths of Western economies also gives benefits to these same countries in institutions that rely
on economic size in their determination of votes. Of course, this technical issue pales in comparison to
the larger problem of assigning voting power to countries based on any measure of output (Figure 14).
 NEW POLITICAL ECONOMY 17

Figure 13. Per capita Convergence of BRICs to US. Source: Authors’ calculations based on GDP shares from the United Nations
Main Aggregates Database (https://unstats.un.org/unsd/snaama/dnlList.asp) and GDP in $PPP from the World Bank.

Figure 14. Alternative Shares of Global Economy, GDP vs. NMVA. Source: Authors’ calculations based on GDP shares from the
United Nations Main Aggregates Database (https://unstats.un.org/unsd/snaama/dnlList.asp) and GDP in $PPP from the World
Bank.
18 J. ASSA AND I. H. KVANGRAVEN

Changes in national accounting standards also have serious implications for admittance criteria to
international organisations such as the EU. For example, the SNA 93 revision made the Greek gov-
ernment deficit look smaller, by inflating the denominator (GDP) compared to SNA 1968. More
broadly, Greece’s economy has gone through structural changes with an increase in imputed
value-added industries, surpassing measured value-added activities in 2009. As Figure 15 shows,
German GDP received a modest boost of between 2 and 3 per cent when it moved from SNA 68
to SNA 93 and also to SNA 2008. However, the GDP of Greece – recently considered an example
of irresponsible economic policies – increased by nearly a quarter on the eve of adopting the
euro (switching from SNA 68 to SNA 93), and a more modest increase when switching to SNA
2008. While it was the Greek government which was formally challenged for having manipulated
its GDP figures by Eurostat (2004), the effects of this legal SNA change are still an illustrative
example of how the SNA reforms can have important political economy implications.
The boost to Greek GDP in the 1993 SNA reflects the increased importance ascribed to certain
sectors in the ‘new economy’ literature, most prominently services and FIRE sectors (note that
retail and wholesale trade also doubled in value under SNA 1993). Figure 16 shows the increases
by industry for 1995, the last year for which SNA 68 data are available. Agriculture, mining, govern-
ment and public services actually decreased in value in the new SNA. While these changes are impor-
tant for the convergence debates as they alter the size of economies based on their economic
structure, this has not been a subject of concern for economists studying convergence to date
(Figure 17).
The rising proportion of imputation has inflated GDP, which is the denominator in the Deficit-to-
GDP ratio, and as a result, Greek deficits look milder under SNA 93 than they would have under the
earlier, SNA 68 (Figure 18).

Figure 15. Percentage Impact of SNA Changes on Greek and German GDP in 1995. Source: United Nations, Main Aggregates and
Detailed Tables (MADT) database, Table 101, http://data.un.org/Data.aspx?d=SNA&f=group_code%3a101 (last accessed March
2020).
 NEW POLITICAL ECONOMY 19

Figure 16. Percentage Change in Greek GDP by Sector from SNA 68–93. Source: United Nations, Main Aggregates and Detailed
Tables (MADT) database, Table 201, http://data.un.org/Data.aspx?d=SNA&f=group_code%3a201 (last accessed March 2020).

Figure 17. Imputed vs. Measured Value-Added as per cent of Greek GDP. Source: Authors’ calculations based on GDP shares from
the United Nations Main Aggregates Database (https://unstats.un.org/unsd/snaama/dnlList.asp) and GDP in $PPP from the World
Bank.
20 J. ASSA AND I. H. KVANGRAVEN

Figure 18. Greek Deficit, per cent of SNA 68 GDP vs. SNA 93 GDP. Source: United Nations, Main Aggregates and Detailed Tables
(MADT) database, Table 202, http://data.un.org/Data.aspx?d=SNA&f=group_code%3a202 (last accessed March 2020).

Taking out all imputations (using NMVA), the picture is even more dramatic, as Figure 19
demonstrates.
It appears likely that, had the EU considered the ratio of government deficit to NMVA rather than
GDP, Greece may not have been admitted to the Eurozone (given the Maastricht Treaty criteria of no
more than 3 per cent Deficit-to-GDP). GDP was similarly important for the calculation of debt relief
for highly indebted poor countries (HIPC), where public debt in relation to GDP and other macroe-
conomic variables had a direct impact on debt relief was granted (Hjertholm 2003).
In addition, GDP growth plays an important role in the Sustainable Development Goals (SDGs), as
over 8 per cent of the indicators rely on GDP (19 of 232). For example, consider Goal 8 to ‘Promote
sustained, inclusive and sustainable economic growth’. Target 8.1 calls for a 7 per cent annual GDP
growth rate per capita in the least developed countries. And while Target 8.2 is about achieving
higher productivity through diversification, technological upgrading and innovation through focus-
ing on high-value added and labour-intensive sectors, the indicator of this target is reduced to GDP
growth per employed person. As there is no obvious connection between the imputation-heavy GDP
and productivity or diversification, and therefore the indicator could be satisfied by routes that
would not satisfy the target (e.g. through growth in the financial sector or increased production
of primary commodities). Although governments in developing countries may not be directly incen-
tivised by SDG goals and targets (see e.g. Horn and Grugel 2018), relying on standard GDP as an indi-
cator for the SDGs is fraught with potential problems. Beyond the methodological problems with
how GDP is measured based on the structure of Western economies, there are also well-documented
problems associated with GDP measurements in developing economies (Dawnson et al. 2001,
Johnson et al. 2009, Ciccone and Jarocinksi 2010, Jerven 2012a, 2012b).
Economists have tirelessly tried to identify what drives economic growth, often by running
regressions with GDP growth – in what Durlauf et al. (2005) refer to as the ‘growth regression indus-
try’. GDP thus also forms the epistemic basis for economic research and policy decisions. Indeed, the
 NEW POLITICAL ECONOMY 21

Figure 19. Greek Deficit, per cent of GDP vs. per cent of NMVA. Source: United Nations, Main Aggregates and Detailed Tables
(MADT) database, Table 202, http://data.un.org/Data.aspx?d=SNA&f=group_code%3a202 (last accessed March 2020).

way growth is measured has important implications on the types of sectors that are considered
growth-enhancing. For example, the IMF (2018) recently argued that the declining share of manu-
facturing jobs need not hurt growth and that services could be considered the new engine of
growth. Such arguments must be theoretically and empirically scrutinised closely, as one can
easily be misled by what drives development by using a biased measure of economic growth.
For countries that wish to demonstrate economic strength, how GDP is measured could have
implications for the kinds of policies governments pursue to increase their GDP. This is in line
with Speich’s (2011) observation that economic statistics are prominent agents of societal change,
as they can shape what kinds of activities come in focus (see also Seers 1972). For example, while
the measure of NMVA as explained above would likely motivate countries to focus their production
on traditional sectors such as manufacturing, utilities, infrastructure and related services, standard
GDP opens up a variety of routes through which countries can increase their GDP, including
through financialisation.
There have been calls for the cultural analysis of the history of economic knowledge to be its own
sub-field of economics, because we often take the epistemic practices of economic knowledge pro-
duction for granted (Schatzki et al. 2001, Schabas 2002, Speich 2011). This history is of course closely
related to national accounting and bookkeeping techniques. While economists and the public at
large tend to see numbers as more precise than qualitative evidence of economic growth, there
are many cases – especially with GDP – where producing one single number to represent economic
output may obscure more than it clarifies (Jerven 2012b).

Conclusion
As with the evolution of economic theory and policy, macroeconomic indicators are fraught with
conceptual and political debate that would benefit from broader debate among social scientists,
22 J. ASSA AND I. H. KVANGRAVEN

and made accessible to the public. Given how central GDP is to economic reporting, economic
research and generally for our perception of growth in the world, understanding how GDP is
measured, how this has changed over time, and the implications of these changes for the conver-
gence debates is crucial. Choices of measurement have fundamental distributive consequences
(Mügge 2016).
Our findings are twofold: (1) the standards of measuring GDP have changed significantly and
through an imperfect technocratic process over the past decades in ways that are not related to
any particular economic theory, but rather that follows structural changes in economies of the
West; and (2) these changes have a substantial impact on the perceived growth rates and relative
strengths of different countries in the world, which in turn has critical political economy implications.
Without the imputed values of some service sectors, the non-Western world has in fact caught up
with the West to a larger extent than what current version of GDP suggests. We therefore argue that
changes in GDP measurement constitute a form of ‘imputing away convergence’, by redefining
growth according to the structure of production of advanced countries. By analysing convergence
with the Narrow Measure Value Added (NMVA) measure, which is closer to previous versions of the
SNA, this paper illustrates that developing economies would have higher growth with older measures
of GDP, and that they are catching up with advanced economies at faster rates with this measure.
The political and economic effects of this development include the justification of developing
economies being given smaller voting shares in international institutions than they would otherwise
be able to demand based on the older measures of GDP, and the new versions of GDP may have
made it possible for countries to join the EU, which otherwise may not have been admitted. At an
epistemological level, countries’ growth fuelled by financialization and real-estate bubbles may
show rapid GDP growth (at least in the short run) without any real structural transformation, thus
giving misleading incentives for other countries to follow suit.
We hope this paper can contribute to an increase in transparency in changes to measures of pro-
duction and their implications and make public the debate about how GDP is measured and
reformed.

Notes
1. This also echoes Jerven’s (2012b) observation that the way we measure economic output tends to be in line with
the Eurocentrism that pervades our field.
2. Ha-Joon Chang (2002) argued that countries in the West used a variety of development policies to get to where
they are, but once there, preached the opposite policies to poorer countries. While we do not claim that devel-
oped countries deliberately manipulate the SNA in order to outpace developing countries, we use this analogy
to illustrate how the changes have in fact moved the developmental goal-post beyond the sectors originally
associated with industrialisation, such as manufacturing, mining, utilities and transport.
3. We are aware of NMVA’s limitations, including its limited acceptance and various critiques. However, the other
option for a counterfactual – using existing series of SNA 1968 data – is severely limited by data availability,
which makes any cross-country comparisons at the aggregate level impossible. We thus use NMVA as an imper-
fect proxy for a measure of value-added which has less imputations than GDP.
4. However, if we exclude China, the ratio per capita income in the West actually increased in the 1900s (Wade
2014).
5. Thus far, the debates in English academic literature on China’s growth have largely been about whether Chinese
growth statistics are plausible or not (e.g. Rawski 2001, Klein and Özmucur 2002, Holz 2004, Mügge 2016), rather
than studying the politics of measurement of China’s economic output on its own terms. This is an important
gap for future research.
6. Such overlap in years for which data is available under multiple SNA systems is limited. This is especially the case
for countries which have never used SNA 1968, including many countries previously dubbed as being ‘in tran-
sition’ from command economies to market-based economies. The list includes the former USSR, other east-
European countries, as well as China and Cuba. Even for non-Communist countries, data is limited, especially
for developing countries.
7. It is important to bear in mind that the difference between any two numbers in the SNA accounts is the com-
bination of changes to the methodology, which is the focus of this paper, and other changes, including improve-
ments in the underlying raw data, changes to the series’ base year, etc. In practice it is difficult to disaggregate all
 NEW POLITICAL ECONOMY 23

these changes, and as a result, comparisons of growth in constant prices in this paper (those using MADT tables
102 and 202 as a source) should be interpreted with caution. This is less of an issue with the current price com-
parisons in this paper (using MADT tables 101 and 201) since there is no change of base year to affect the results.
8. This is based on the (implicit) market-centered assumption that where there is income there must be production.
As Mazzucato (2018) demonstrates, this logic is circular. Income is earned from productive activities, and activi-
ties are considered productive since they earn income. This not only leaves-out unpaid activities, but also lets in
unearned income, or what the classical political economists thought of as rentier income (earned from owner-
ship, control or even monopolistic power).

Disclosure Statement
No potential conflict of interest was reported by the author(s).

Notes on contributors
Jacob Assa holds a Ph.D. in economics from the New School for Social Research, and his dissertation on the financializa-
tion of GDP has been published as a book by Routledge in 2016. He is currently a policy specialist with UNDP, and pre-
viously worked for 14 years in UN-DESA’s statistics division, in the areas of national accounts, development indicators,
and as chief of the statistical dissemination section. Jacob has published in several peer-reviewed journals, and his
research interests include the political economy of national accounting, financialization, inequality and growth, com-
posite indicators, and sustainability.
Ingrid Harvold Kvangraven is a Lecturer in International Development at the University of York, UK. Her research fields
include development finance, African debt markets, heterodox economics, decolonizing economics, international insti-
tutions, and history of economic thought. She holds a PhD in Economics from The New School for Social Research.

References
Assa, Jacob, 2015. Gross domestic power: geopolitical economy and the history of national accounts. In: R. Desai and P.
Zarembka, eds. Theoretical engagements in geopolitical economy: part A. Bingley, UK: Emerald Group, 175–203.
Assa, Jacob, 2017. The financialization of GDP: implications for economic theory and policy. New York: Routledge
Advances in Heterodox Economics.
Assa, Jacob, 2018. Finance, social value, and the rhetoric of GDP. Finance and Society, 4 (2), 144–158, November.
Barro, Robert J., 1991. Economic growth in a cross section of countries. Quarterly journal of economics, 106, 407–443.
Barro, Robert J. and Sala-i-Martin, Xavier, 1992. Convergence. Journal of political economy, 100, 223–251.
Basu, Deepankar and Foley, Duncan, 2013. Dynamics of output and employment in the US economy. Cambridge journal
of economics, 37 (5), 1077–1106.
Baumol, William J., 1986. Productivity growth, convergence and welfare: what the long run data show? American econ-
omic review, 76, 1072–1085.
Bos, Frits. 1995. Economic theory and national accounting. Statistics Netherlands Working Paper NA-075.
Chang, Ha-Joon, 2002. Kicking away the ladder: development strategy in historical perspective. London: Anthem Press.
Christophers, Brett, 2011. Making finance productive. Economy and Society, 40 (1), 112–140. doi:10.1080/03085147.2011.
529337.
Ciccone, Antonio and Jarocinksi, Mark, 2010. Determinants of economic growth: will data tell? American Economic
journal: macroeconomics, 2 (4), 223–247.
Cooley, Alexander and Snyder, Jack, 2015. Ranking the world. Grading states as a tool of global governance. New York:
Cambridge University Press.
Corrado, Carol, Haskel, Jonathan, and Jona-Lasinio, Cecilia, 2017. Public intangibles: the public sector and economic
growth in the SNA. The review of income and wealth, 63 (2), 355–380.
Coyle, Diane, 2014. GDP: a brief but affectionate history. Princeton, NJ: Princeton University Press.
Davis, Kevin, et al., 2012. Governance by indicators. global power through classification and rankings. New York: Oxford
University Press.
Dawnson, John W., et al., 2001. Economic information versus quality variation in cross-country data. Canadian journal of
economics, 34 (3), 998–1009.
Decancq, Koen and Schokkaert, Erik, 2016. Beyond GDP: using equivalent incomes to measure well-being in Europe.
Social Indicators Research, 126 (1), 21–55.
Delong, Bradford, 1988. Productivity growth, convergence, and welfare: comment. American economic review, 78 (5),
1138–1154.
24 J. ASSA AND I. H. KVANGRAVEN

Durlauf, Steven, Johnson, Paul, and Temple, Jonathan, 2005. Growth econometrics. In: Philippe Aghion and Steven
Durlauf, eds. Handbook of economic growth. Amsterdam: Elsevier, 555–667.
Eurostat, 2004. Report by Eurostat on the revision of the greek government deficit and figures. Luxembourg: The European
Commission.
Feldstein, Martin, 2017. Underestimating the real growth of GDP, personal income, and productivity. Journal of econ-
omic perspectives, 31 (2), 145–164.
Felice, Emanuele, 2015. The misty Grail: The search for a comprehensive measure of development and the reasons for
GDP primacy. Munich personal RePEc archive paper, 61095.
Fioramonti, Lorenzo, 2013. Gross domestic problem. The politics behind the world’s most powerful number. London: Zed
Books.
Fleurbaey, Marc and Blanchet, Didier, 2013. Beyond GDP. Oxford: Oxford University Press.
Folbre, Nancy, 2015. Valuing non-market work. UNDP Human Development Report Office. Available from: http://hdr.
undp.org/sites/default/files/folbre_hdr_2015_final_0.pdf [Accessed 15 June 2018].
Foley, Duncan, 2013. Rethinking financial Capitalism and the ‘information economy’. Review of radical political econ-
omics, 45 (3), 257–268.
Fougner, Tore, 2008. Neoliberal governance of states: the role of competitiveness indexing and country benchmarking.
Millennium, 37 (2), 303–326.
Fraumeni, Barbara, Christian, Michael, and Samuels, Jon, 2017. The accumulation of human and nonhuman capital,
revisited. The review of income and wealth, 63 (2), 381–410.
Grier, Kevin B. and Tullock, Gordon, 1989. An empirical analysis of cross-national economic growth, 1951–1980. Journal
of monetary economics, 24, 259–276.
Griliches, Zvi, 1994. Productivity, R&D, and the data constraint. The American economic review, 84 (1), 1–23.
Hansen, Krause H. and Muhlen-Schulte, A., 2012. The power of numbers in global governance. Journal of international
relations and development, 15 (4), 455–465.
Hansen, Krause H. and Porter, T., 2012. What do numbers do in transnational governance? International Political soci-
ology, 6 (4), 409–426.
Hickel, Jason, 2019. The sustainable development index: measuring the ecological efficiency of human development in
the Anthropocene. Ecological Economics, 167, 106331.
Hjertholm, Peter, 2003. Theoretical and empirical foundations of HIPC debt sustainability targets. Journal of development
studies, 39 (6), 67–100.
Holtz-Eakin, Douglas, 1993. Solow and the states: capital accumulation, productivity, and economic growth. National tax
journal, 46, 425–439.
Holz, Carsten A, 2004. Deconstructing China’s GDP statistics. China economic review, 15, 164–202.
Horn, Philipp and Grugel, Jean, 2018. The SDGs in middle-income countries: setting or serving domestic development
agendas? Evidence from Ecuador. World development, 109 (C), 73–84.
Hulten, Charles and Nakamura, Leonard, 2020. Expanded GDP for welfare measurement in the twenty-first century,
NBER Working Paper Series 26578.
IMF, 2018. World economic outlook, April 2018 cyclical upswing, structural change. Washington, DC: International
Monetary Fund.
Irvine, John, Miles, Ian, and Evans, Jeff, 1979. Demystifying social statistics. London: Pluto Press.
Islam, Nazrul, 2003. What have we learnt from the convergence debate? Journal of Economic Surveys, 17 (3), 309–362.
Jerven, Morten, 2012a. Briefing: for richer, for poorer: GDP revisions and Africa’s statistical tragedy. African Affairs, 112/
446, 138–147.
Jerven, Morten, 2012b. An unlevel playing field: national income estimates and reciprocal comparison in global econ-
omic history. Journal of global history, 7, 107–128.
Jerven, Morten, 2013. Poor numbers. How we are misled by African development statistics and what to do about it. Ithaca,
NY: Cornell University Press.
Johnson, Simon, et al., 2009. Is newer better? Penn world table revisions and their impact on growth estimates. National
bureau of economic research working paper, 15455.
Johnson, Paul and Papageorgiou, Chris, 2020. What remains of cross-country convergence? Journal of economic litera-
ture, 58 (1), 129–175.
Jones, Charles and Klenow, Peter, 2016. Beyond GDP? welfare across countries and time. American economic review, 106
(9), 2426–2457.
Kelley, Judith G. and Simmons, Beth A., 2015. Politics by number. Indicators as social pressure in international relations.
American journal of political science, 59 (1), 55–70.
Kendrick, John, 1970. The historical development of national accounts. History of political economy, 2, 284–315.
Keynes, John Maynard, 1940. How to pay for the war: a radical plan for the chancellor of the exchequer. London: Macmillan
and Co., Ltd.
Klein, Lawrence R. and Özmucur, Süleyman, 2002. The estimation of China’s economic growth rate. Journal of economic
and social measurement, 28, 187–202.
 NEW POLITICAL ECONOMY 25

Kormendi, Roger C. and Meguire, Philip G., 1985. Macroeconomic determinants of growth: Cross-country evidence.
Journal of Monetary Economics, 16, 141–163.
Lepenies, Philipp, 2016. The power of a single number: a political history of GDP. New York: Columbia University Press.
Lequiller, François and Blades, Derek, 2006. Understanding national accounts. Paris: OECD.
Lin, Justin Y. and Rosenblatt, David, 2012. Shifting patterns of economic growth and rethinking development. Washington,
DC: World Bank.
Lucas, Robert, 1990. Why doesn’t capital Flow from rich to poor countries? The American economic review, 80 (2), Papers
and Proceedings of the Hundred and Second Annual Meeting of the American Economic Association, 92–96.
Maddison Project. 2013. The Maddison-Project. Available from: https://www.rug.nl/ggdc/historicaldevelopment/
maddison/releases/maddison-project-database-2013
Mankiw, N. Gregory, Romer, David, and Weil, David, 1992. A contribution to the empirics of economic growth. Quarterly
journal of economics, 107, 407–437.
Masood, Ehsan, 2016. The great invention: the story of GDP and the making and unmaking of the modern world. New York
and London: Pegasus Books.
Mazumdar, Debashis, 2016. The problems of development gap between developed and developing nations: Is there
any sign of convergence? In: Ramesh Chandra Das, ed. Handbook of research on global indicators of economic and
political convergence. Hershey, PA: IGI Global, 29–50.
Mazzucato, M., 2018. The value of everything – making and taking in the global economy. London: Penguin.
Mitra-Khan, B.H., 2011. Redefining the economy: how the ‘economy’ was invented 1620. Unpublished doctoral thesis, City
University London. Available from: http://openaccess.city.ac.uk/1276 [accessed 9 February 2018].
Mügge, Daniel, 2016. Studying macroeconomic indicators as powerful ideas. Journal of European public policy, 23 (3),
410–427.
Mügge, Daniel, 2020. Economic statistics as political artefacts. Review of international political economy. doi:10.1080/
09692290.2020.1828141.
Nayyar, Deepak, ed., 2013. Catch up: developing countries in the world economy. New Delhi: Oxford University Press.
Ocampo, J.A. and Jomo, K.S. 2007. Growth divergences: explaining differences in economic performance. Hyderabad:
Orient Longman.
Perlman, Mark and Marietta, Morgan, 2005. The politics of social accounting: public goals and the evolution of the national
accounts in Germany, the United Kingdom and the United States. Review of political economy, 17 (2), 211–230.
Philipsen, Dirk, 2015. The little Big number: How GDP came to rule the world and what to do about it. Princeton, NJ:
Princeton University Press.
Popov, Vladimir and Jomo, K.S., 2018. Are developing countries catching up? Cambridge journal of economics, 42, 33–46.
Pritchett, Lant, 1997. Divergence, big time. Journal of economic perspectives, 11 (3), 3–17.
Rawski, Thomas, 2001. What is happening to China’s GDP statistics? China economic review, 12, 347–354.
Rebelo, Sergio, 1991. Long run policy analysis and long run growth. Journal of political economy, 94, 1002–1037.
Romer, Paul, 1986. Increasing returns and long run growth. Journal of political economy, 99, 500–521.
Schabas, Margaret, 2002. Coming together: history of economics as a history of science. History of political economy,
Ann. Suppl, 34, 208–225.
Schatzki, Theodore R., Cetina, Karin Knorr, and Savigny, Eike von, eds., 2001. The practice turn in contemporary theory.
New York: Routledge.
Seers, Dudley, 1972. What are we trying to measure? The journal of development studies, 8 (3), 21–36.
Solow, Robert M., 1956. A contribution to the theory of economic growth. Quarterly journal of economics, 70, 65–94.
Speich, Daniel, 2011. The use of global abstractions: national income accounting in the period of imperial decline.
Journal of global history, 6, 7–28.
Stiglitz, Joseph, Sen, Amartya, and Fitoussi, Jean-Paul, 2010. Report by the commission on the measurement of economic
performance and social progress. Paris: INSEE.
Studenski, Paul, 1958. The income of nations. Washington Square, New York: New York University Press.
Swan, Trevor W, 1956. Economic growth and capital accumulation. Economic record, 32 (63), 334–361.
Tavernier, J., Cuneo, P., and Plateau, C., 2015. Measurement of quality of life and well-being in France: the drivers of
subjective well-being. Review of income and wealth, 61 (1), 25–33.
Tily, Geoff, 2009. John Maynard Keynes and the development of national accounts in Britain, 1895–1941. Review of
income and wealth, 55 (2), 331–359.
Toma, Sorin-George, Marinescu, Paul, and Gradinaru, Catalin, 2017. In search of the global economy supremacy: China
versus USA. Annals – economy series, 1, 227–233.
UN, 2009. System of national accounts 2008. New York, USA. https://unstats.un.org/unsd/nationalaccount/docs/sna2008.
pdf.
van der Bergh, J.C.J.M. and Verbruggen, H., 1999. The GDP paradox. Journal of economic psychology, 30 (2), 117–135.
Vanoli, A., 2005. A history of national accounting. Amsterdam: IOS Press.
Wade, Robert, 2013. The art of power maintenance how western states keep the lead in global organizations. Challenge,
56 (1), 5–39.
26 J. ASSA AND I. H. KVANGRAVEN

Wade, Robert, 2014. Growth, inequality and poverty: evidence, arguments, and economists. In: Ravenhill, ed. Global pol-
itical economy (4th ed.).
Ward, Michael, 2004. Quantifying the world: UN ideas and statistics. New York: Indiana University Press.
Waring, Marilyn, 1990. If women counted: a new feminist economics. San Francisco: Harper Collins Publishers.
Wesselink, Bart, et al., 2007. Measurement beyond GDP – background paper. In: Presented at conference measuring pro-
gress, true wealth and the well-being of nations, 19 & 20 November 2007, Brussels, Belgium.
Woods, Dawayne, 2014. The use, abuse and omertà on the ‘noise’ in the data: African democratisation, development
and growth. Canadian journal of development studies, 35 (1), 120–135.

Appendices
Annex I: Measuring Productivity in the Financial Sector
As shown in Assa (2015, 2017), the FIRE sector in most OECD countries has no correlation (or in some cases a negative
correlation) between its share of total value-added in the economy and its share of total employment, while most others
sector have a positive correlation. The Goldman Sachs view mentioned above would interpret this anomaly as evidence
of significant productivity increases in the financial sector, enabling it to produce far more output with less labour input.
But this negative correlation could also be a symptom of a problem with the measurement of output for this sector. This
is explained formally in Assa (2015):
Ni
. Sector i’s share of employment is sei = , where Ni is employment in sector I and N is the total employment in the
N
economy
y Yi
. Sector i’s share of output is si = , where Yi is output in sector i and Y is total output in the economy
Y
Yi
. Sector i’s productivity is Pi =
Ni
Y
. Average productivity in the economy is P =
N
If all sectors of the economy are productive, the relationship between sectoral employment shares and output
shares would be mediated by each sector’s labour productivity relative to the average productivity:
Pi
sei = syi (A1)
P
y
where sei is sector i’s share of employment, pis average productivity, pi is sector i’s productivity, and si is sector i’s share
of output. This can be rearranged as follows:
P
sei = syi (A2)
Pi

Thus, a sector’s share of total employment should be related to its share of output through its productivity relative
Pi
to the average productivity in the economy. Since the latter can never be negative (i.e. ≫ 0), there is indeed a
P
problem with imputing superior productivity to the financial sector given the observed negative correlation
between its share of value added and its share of employment.
Therefore, the reported productivity of the financial sector is more of a statistical artefact than a real phenomenon.
To see why, rearranging (3) gives:
P
Pi = syi (A3)
sei
y
Since si is increasing but sei is decreasing faster for the financial sector, its productivity based on standard national
accounting is indeed too good to be true.

Annex II: Share of Imputed Sectors in GDP

Table A1. Share of imputed sectors in GDP by income group (unweighted average).
1970 2017
Low income 18.5 24.7
Lower middle income 26.6 29.9
Upper middle income 28.7 35.8
High income 31.6 45.9
 NEW POLITICAL ECONOMY 27

Annex III: Additional Figures

Figure A1. Military Expenditure by Income Group (billions of 2011 PPP$). Source: Authors’ calculations based on World Bank,
World Development Indicators database, series MS.MIL.XPND.GD.ZS. Accessed March 2020.

Figure A2. Military Expenditure by Region (billions of 2011 PPP$). Source: Authors’ calculations based on World Bank, World
Development Indicators database, series MS.MIL.XPND.GD.ZS. Accessed March 2020.
28 J. ASSA AND I. H. KVANGRAVEN

Figure A3. Military Expenditure of China and Japan (billions of 2011 PPP dollars, Constant Prices). Source: Authors’ calculations
based on World Bank, World Development Indicators database, series MS.MIL.XPND.GD.ZS. Accessed March 2020.

Figure A4. R&D as a per cent of GDP by Region. Source: Authors’ calculations based on World Bank, World Development Indi-
cators database, series GB.XPD.RSDV.GD.ZS. Accessed March 2020.
 NEW POLITICAL ECONOMY 29

Figure A5. R&D in constant prices (billions of $PPP) by Region. Source: Authors’ calculations based on World Bank, World Devel-
opment Indicators database, series GB.XPD.RSDV.GD.ZS. Accessed March 2020.

Figure A6. R&D Expenditure as per cent of GDP vs. log of Per Capita Income, 2013–2017 averages. Source: Authors’ calculations
based on World Bank, World Development Indicators database, series GB.XPD.RSDV.GD.ZS and NY.GDP.PCAP.PP.KD. Accessed
March 2020.
30 J. ASSA AND I. H. KVANGRAVEN

Annex IV: Convergence Regressions using Market Exchange Rates

Table A2. Unconditional convergence 1970–2017.
GDP NMVA
Log per capita 1970 −0.234 −0.379
significance *** ***
R2 0.039 0.110

Table A3. Conditional convergence 1970–2017.

GDP NMVA
PopGr −0.736*** −0.625***
AvgSav7017 0.019 0.017
AvgGCF7017 0.081*** 0.090***
lnPop1970 0.068 0.072
AvgPrim7017 0.009 0.010
Log per capita 1970 −0.486*** −0.611***
R2 0.417 0.472