‘Not too high, not too low’: transparency, opacity and the politics of poverty measurement in Jordan

ABSTRACT

This paper explores the politics of creating and calibrating monetary poverty indicators in Jordan using interviews with policy-shapers and documentary analysis. It highlights the significance of these dynamics for conceptualizing governance and statehood in the Middle East. I argue that poverty indicators have served a dual purpose: they have functioned as a tool of state legibility, seeking to enable governments to act on poverty and increase accountability. At the same time, opacity in their production has made it possible to shirk responsibility for worsening socio-economic situations. The combination has helped to reproduce the state as a distinct entity that should, at least in principle, be able to tackle socio-economic inequalities. By empirically and conceptually highlighting the intertwinement between transparency and opacity, the article not only contributes a new perspective to debates around governance through indicators, but also to de-exceptionalizing the Middle East in discussions on the globalized politics of development.

KEYWORDS:

• Indicators
• poverty
• opacity
• government
• Jordan
• Middle East

Poverty indicators, especially monetary ones, are a globalized, standardized yet highly contentious affair. Consider Jordan. Donors and international agencies have long praised it as one of the few countries in the region that produce reliable socio-economic indicators (e.g. Adams & Page, 2003; Shaban et al., 2001, p. 4). Since it standardized its poverty measurements in the early 2000s, the absolute national poverty rate has fluctuated between 11% and 16% of the overall population – a not irrelevant but seemingly manageable level. At the same time, international and Jordanian observers alike have raised doubts about the validity of these indicators (e.g. Harrigan et al., 2006a). The Government of Jordan (GoJ) has also not published a poverty report since 2012 that fully details the state of poverty in the country or how the figures are calculated. This has drawn strong criticism from observers, especially from within Jordan. They assume poverty rates have risen sharply in the past decade due to the ongoing economic crisis, IMF austerity policies, and lasting repercussions from the Syrian conflict (e.g. ARDD, 2019; Bibars, 2019).

What does this tension tell us about politics, the economy, and the state in Jordan? Globalized indicators in this regional context have received limited scrutiny to date (cf. see Baumann, 2017; Bourmaud, 2011; Destremau, 2001; Rignall & Atia, 2017); however, their creation and calibration can serve as a window onto the political economy of Jordan and the Middle East. In recent years, a range of scholars have analysed the ways in which government has been transformed through a variety of agencies, non-human actants and forms of expertise that cross public and private as well as national and international lines (e.g. Debruyne, 2013; Parker, 2009; Schütze, 2019). Rather than exceptionalizing Middle Eastern political economies this new literature has focused on understanding the regulation of economic and social issues in practice, in specific locations, and regarding particular populations (e.g. Abu-Hamdi, 2017; Lenner, 2015; Martínez, 2022). This has come with a conceptual openness to both the strength and failures of neoliberal governmentality, and what those mean for the reproduction of the state. Martínez (2022) elucidates how the state in Jordan emerges through a variety of governmental techniques for regulating the socio-economic life of populations. Focusing on bread subsidies, he shows how this works despite and even through failure. Bogaert (2018) and Schütze (2019) highlight how globalized projects of improvement, for example democracy promotion and human development initiatives, reproduce the problems they have sought to overcome, thereby strengthening authoritarian power.

By exploring the politics of producing and calibrating poverty indicators, this article offers ambiguity and opacity as a conceptual lens for furthering this strain of political-economic research, and for advancing scholarship on the governmentality of indicators. I show how the creation and dissemination of poverty indicators in Jordan serve a dual purpose. As tools of state legibility, these indicators enable governments to act on poverty and increase accountability. However, opacity around their production can also obscure worsening socio-economic situations, and deflect responsibility for the situation away from government offices. I argue that this combination of rendering poverty legible while also obscuring it has helped reproduce the state as an entity that should, in principle, be able to tackle socio-economic inequalities. This has reinforced authoritarian power during a period of structural adjustment and decreasing social welfare allocations, in which targeted poverty alleviation largely displaced the project of the state as a provider of social welfare, and in which poverty indicators became a new yardstick for state performance.

To develop this argument, I first bring into dialogue discussions on the role of indicators in global governance, debates on poverty measurement, and scholarship on ambiguity and opacity in political life to draw out the ways in which this sheds light on the enactment and reproduction of statehood. I then introduce my case study by tracing the development of poverty indicators in Jordan. These first emerged as a tool of legibility in Jordan during the 1980s and then underwent a process of standardization as they became central to the politics of aid and structural adjustment. Subsequently, I explore the methodological-political debates around money-metric poverty measurements. I show how recalculations and changes in methodology have consistently resulted in poverty rates that are not too high, and not too low. Not only are these moderate poverty rates palatable for many audiences, but they are also very hard to contest. Finally, I reflect on the role of transparency and opacity in creating the representation of a state that is responsible for and principally capable of reducing socio-economic inequalities.

This article is based on over 30 semi-structured interviews and informal conversations with donor representatives, statisticians, ministry officials, consultants, and diverse poverty experts, which I conducted in Amman, Jordan in English and Arabic between 2006 and 2016 as part of a broader project on the politics of poverty alleviation policy. This is combined with documentary analysis of (official and unofficial) poverty reports in English and Arabic from 1988 to 2020, which I obtained from interview partners’ private archives, as well as Arabic-language newspaper and journal articles commenting on them.

Opacity and ambiguity in the politics of governing through indicators

Over the course of the last decade, scholarship on globalized indicators as a technology of government has exploded. Inspired particularly by science and technology studies and governmentality studies, many contributions have investigated how socio-economic and governance indicators have reshaped the conduct of governments and administrations, and of the populations identified as their objects (e.g. Davis et al., 2012a; Rottenburg et al., 2015).

Two aspects are particularly relevant for the purposes of this article. First, scholars have highlighted how indicators shape and often even create the fields of intervention they are supposed to describe. As tools of legibility, which seek to visibilize, categorize, and standardize populations according to discrete biopolitical issues, they create the possibility for action as well as for being seen to be acting (Rose, 1999; Scott, 1998). At the same time, they tend to depoliticize these same issues by congealing them into numbers (Davis et al., 2012b; Mitchell, 2002).

Second, the process of governing through indicators is analysed as part of what creates the effect of states standing apart from international agencies and dynamics, as well as the populations they seek to govern. By facilitating transnational standardization and comparison in relation to prescriptive policies designed by international institutions, they help with these agencies’ attempts to govern from a distance (e.g. Mitchell, 2002; Rottenburg & Merry, 2015). Löwenheim, for example, argues that they produce an ‘examined state’ as ‘an actor deemed capable of action and responsible for choices and policies – and, consequently, for the scores it receives’ (Löwenheim, 2008, p. 259). Thereby, the global politics involved in creating the problems the indicators seem to merely depict – poverty, corruption, or health – become relatively invisible, and states move into the foreground as those accountable for remedying them. As I will argue in this paper, this invisibilization attempt does not always succeed, yet poverty indicators still contribute to producing the effect of a responsible state in principle.

As Mitchell elaborates, this effect partly comes about due to the transfer of knowledge creation from specific locales and social relationships to statistical agencies located in administrative centres, and dominated by experts. This process is part and parcel of the emergence of a seemingly free-standing national economy; a supposedly distinct social sphere and object of centralized governmental regulation (Mitchell, 2002, pp. 100–103). As the Jordanian example shows, the centralization of data collection and aggregation within one single agency not only contributes to this effect. It also severely limits the ability of ‘other’ experts to generate alternative understandings of the socio-economic world.

Another relevant strand of discussion revolves around indicators’ role in fostering trust in government, making it seem accountable, rational, effective or fair (e.g. Espeland & Stevens, 2008; Porter, 1995). As Rottenburg and Merry (2015, p. 23) put it, ‘[i]ndicators are there to verify that it is safe to put one’s lives and hopes into the hands of the state apparatus.’ Of course, in practice, this does not always work. Indicators can also be severely mistrusted, or turned against governmental (in)action by scandalizing inequalities and discrimination (see Rottenburg & Merry, 2015).

To better understand precisely how indicators buttress or destabilize governmental rule, it is vital to devote more attention to the practices, rationales, and effects of specific methodologies for calculating indicators (see Baumann, 2019, pp. 507–508). With regard to poverty indicators, there is a relative disconnect between debates on different methodologies for calculating poverty (usually among development economists working with or close to international development institutions), and analyses of the effects of governing through these indicators, as in the scholarship mentioned above. The former often have a clear awareness that methodological choices, for example, around ways to calculate income and expenditure, or for setting the poverty line, are not only relatively arbitrary but also have profound consequences for the resulting indicators. They acknowledge, for example, how differences in setting minimum caloric requirements, the composition of a food basket, the calculation of a price index or different definitions of income can have a dramatic impact on who is defined as poor, or on the scale of inequality (see e.g. Ravallion, 2010; Rose, 2018). The moral and political rationales feeding into these choices, as well as their socio-economic and political consequences, remain under-analysed however. On the other hand, scholarship on socio-economic indicators as a technology of governance has not delved deeply into methodological issues.

An important exception is Fischer’s (2018) in-depth analysis of the political and normative choices feeding into and effects of various methods and measures of poverty and inequality. In order to dig deeper into the governance effects of methodological choices, I focus on the significance of ambiguity and opacity in setting and calibrating money-metric poverty lines and rates.

The role of ambiguity and opacity in governance processes has attracted increased scholarly attention vis-à-vis a variety of fields, ranging from (forced) migration control (Stel, 2020; Tazzioli, 2021) and corporate and organizational management (Birchall, 2011; Cappellaro et al., 2021; Davenport & Leitch, 2005) to the politics of international development (Bergamaschi, 2009; Best, 2012) or the functioning of authoritarian government (Wedeen, 1999). At the heart of this scholarship is the assumption that ambiguity constitutes a normal aspect of life, but that it can also be purposefully produced. While bureaucratic organizations systematically seek to reduce ambiguity, they also rule with and through ambiguity (Best, 2012; Stel, 2020). This can involve the production of opacity, i.e. the reduction of transparency. Purposefulness should not be conflated with intentionality though. As Tazzioli (2021, p. 7) highlights: ‘Opacity is not necessarily the outcome of a coherent strategy of obfuscation, nor does it consist in acts of deliberate concealment.’ It can as well be the contingent outcome of heterogeneous strategies and interests.

Ambiguity and opacity can be productive, as vagueness allows governmental actors to navigate interests and maintain flexibility in changing contexts. This, in turn, can help mobilize sufficient support for specific policies or enable organizational change (Davenport & Leitch, 2005). Cappellaro et al. (2021, p. 1) also highlight the protective function of ambiguity, and particularly opacity, by explaining how they ‘allow organizations to conceal themselves from public scrutiny and negative evaluations by external actors’. This, in turn, can contribute to shirking responsibility for socio-economic or political grievances (Davenport & Leitch, 2005).

Another strand of scholarship highlights the disorienting and demotivating effect of institutional ambiguity and opacity. This can relate to particularly vulnerable populations such as (forced) migrants (e.g. Stel, 2020; Tazzioli, 2021) but can also affect populations at large, for example, in authoritarian settings like Syria or Iraq under Saddam Hussein, in which personality cults have served as a mechanism of rule (Wedeen, 1999). If regulations and political messaging are blurry, unpredictable, overlapping, or subject to frequent changes, this can drain agency and keep populations from claiming rights – thereby facilitating their continued demobilization, marginalization and exploitation.

Much of the reviewed scholarship contrasts opacity with transparency, yet a few papers productively highlight the importance of their complementarity. Birchall (2011, p. 14), for example, describes ‘countless ways for corporations and financial institutions to appear transparent to outside observers but keep certain practices opaque, including creative accounting, the release of data incomprehensible to the lay investor or customer, and the hounding of whistleblowers.’ Indeed, in the creation and calibration of poverty indicators, transparency and opacity often go hand in hand.

I use these broader arguments on ambiguity and opacity to extend the literature that points to their significance in the process of governing through socio-economic indicators (e.g. Porter, 2015, p. 54). Fischer (2018, p. 65, 106) mentions the significance of opacity in money-metric poverty measurements and explains that choosing non-intuitive measures (such as proxy consumption measures) over more accessible national poverty rates expressed in local currencies, can disorient and thereby prevent overt contestation by local populations. I will show in the following that obfuscation is equally crucial in setting and calibrating national poverty rates, particularly because they are relatively accessible as such.

Bergamaschi (2009) argues that a poverty reduction strategy matrix in Mali, notable for its vague and sometimes outright irrelevant data, enabled both donors and government to interpret the data in flexible ways and even to jointly manipulate them. This serves the interests of both sides, but ultimately strengthens the sovereignty of the Malian state. While highlighting the role of opacity, Bergamaschi omits the significance of transparency and dismisses the function of indicators as a tool of legibility. This is, she argues, because the main purpose of poverty indicators is to suit donor demands rather than to guide interventions (Bergamaschi, 2009, pp. 134–135).

The Jordanian case shows, though, that indicators can serve both functions simultaneously. They can be a legitimating tool in the politics of aid and serve as tool of legibility and basis for governmental action. In combining transparency and opacity, I argue, the methodologies chosen have yielded poverty rates that are ‘not too high and not too low’ – a desirable outcome for donors, international organizations and Jordanian governmental agencies alike. This has contributed to insulating the process from critique through both experts and non-experts, and to producing the effect of the state as a coherent actor distinct from society and economy that is, or at least should be, capable of acting upon them (Mitchell, 1991).

Making poverty legible in Jordan

Poverty indicators function as a tool of legibility. They help make visible social problems that are seen as potential threats to the political establishment and the broader social fabric, which in turn can inform specific forms of intervention (Destremau, 2001; Lautier, 2013). For aid-receiving states like Jordan, poverty indicators are also deeply entangled with the politics of aid. Although donor assistance to MENA countries depends much more on political factors than on poverty levels (Harrigan et al., 2006b), they still feed into strategies for negotiating aid packages and serve to legitimate decisions about aid relationships and funding streams. In this way, they signal not only the ability of states to act on the issue of poverty but also of the donor-funded strategies behind them. They function as a commentary on globalized policies, or the lack of appropriate or effective interventions. In the period of the post-Washington consensus, in which structural adjustment and the retrenchment of welfare provision were accompanied by targeted poverty alleviation programmes, poverty indicators have often served as a yardstick for the success or failure of these policies (e.g. WB, 1994).

Current measures of poverty and inequality used across countries of the global South, including Jordan, involve clear North–South hierarchies of knowledge. They are European ways of conceptualizing poverty that were made mobile and brought to bear on previously colonized countries via international organizations (e.g. Bonnecasse, 2011). This goes both for the World Bank’s poverty line approach and alternative approaches, such as UNDP’s Multidimensional Poverty Index (MPI). The World Bank assumes that poverty constitutes the inability of an individual to financially meet their basic human needs. Accordingly, it measures the minimum per capita expenditure needed in a given context and calls that the ‘poverty line’. In the perspective underlying the MPI, on the other hand, basic needs do not constitute an end in themselves. They are only a means to fulfil other necessities of life and widen choices, which means that poverty cannot be summarized in a monetary measure alone (cf. Destremau, 2001, p. 132).

While competing forms of measuring poverty have different histories and political implications (see Hacking, 2000; Ward, 2004), the singling out of particular populations and putting numbers on their plight, whether monetary or otherwise, has a distinct effect. It individualizes that problem and helps devise technical, depoliticized solutions to it, rather than centring issues of social justice, rights and citizenship. Poverty threshold logics, whether money-metric or multidimensional, sideline post-war attempts across the world to transform societies through comprehensive processes of redistribution (Fischer, 2018, pp. 31–45; Lautier, 2013). This is compounded by the lasting dynamics of postcolonial aid relationships, which continue to single out poor populations as ‘others’ whose problems can be fixed without tackling the broader inequalities caused by global capitalism and the legacies of colonial rule (e.g. Roy, 2016, pp. 32–37). More often than not, these forms of measurement, and the narratives and solutions accompanying them, are based on standardized formats meant to enable cross-country comparison. As such, they are tailored more to the needs of international organizations than to those of the recipient countries (e.g. Mitchell, 2002, pp. 209–243).

The emergence of poverty measurements

Despite these knowledge and funding hierarchies, the history of conceptualizing and measuring poverty in Jordan shows that globalized approaches are not translated in a straightforward way into the context of countries in the global South. By the early 1970s, poverty had already moved front and centre in international development institutions under Robert McNamara’s reign as World Bank director. Absolute income poverty was a settled unit of analysis, and poverty was conceived as a property of specific – primarily rural – populations that needed to be targeted to satisfy their basic needs (Pereira, 2020). Language hinting at these global discourses was also used in Jordan’s earliest development plans.¹ Yet Jordan’s focus during the 1970s and early 1980s was on industrialization and raising the population’s income levels through governmental transfers, rather than on poverty proper or rural development projects. GDP per capita mixed with measures of government expenditure, rather than poverty measures, were used to assess living standards (Bonnecasse, 2011; Harrigan et al., 2006a, p. 281).

The push for an expansion of planning capacity, including socio-economic data gathering, first began in the 1960s, triggered by national and regional political tensions in 1957/8 that threatened to destabilize the rule of Jordan’s young new monarch, King Husayn (Schayegh, 2013). Population surveying to protect populations from price rises was significantly expanded after an army mutiny in 1974, followed by a major expansion of social provision and the introduction of price controls and universal subsidies, which signalled a much more interventionist form of government (Martínez, 2022, pp. 34–49). Attempts to make the socio-economic situation legible for the sake of governmental intervention were thus clearly driven by domestic and regional events. They were not, however, independent from globalized expertise.

A similar dynamic was at work when poverty indicators began to be properly tabulated. In many aid-receiving countries, including in the Middle East, this only took off as a correlate of structural adjustment policies. These ‘poverty assessments’ dominated knowledge production and indicator creation in the 1990s and sought to enable cross-country comparison. They were largely written by a mix of World Bank staff and foreign and local consultants and provided bundled, country-specific poverty profiles and national poverty rates based on quantitative analyses of national household survey data. In addition to data, they gave policy recommendations that followed the World Bank’s standard approach at the time (Hanmer et al., 1999).

In Jordan though, poverty reporting and the development of poverty indicators started before the onset of structural adjustment, and without the direct involvement of international organizations. It was a corollary of falling oil prices, which led to reduced demand for labour migration to the Gulf, as well as falling aid levels. In the late 1980s, these developments culminated in a debt crisis, rapid currency devaluation, high unemployment, and decreased living standards (cf. Satloff, 1994). This increased public and institutional awareness of poverty as a nation-wide problem that needed to be tackled, rather than only one of specific (and often marginalized) population groups, and triggered the first attempts to design and calculate proper poverty indicators that would make the phenomenon legible (Lenner, 2014).

The first studies were conducted at the request of King Husayn by a fully Jordanian team comprising the then general secretary of the Ministry of Social Development (MoSD) and a group of researchers at a Jordanian university (Abū Khalīl, 2005; Abū Khalīl, 2019). With government funding for an in-depth study, the team compiled two Arabic-language reports (MoSD, 1989, 1993). The first set the groundwork for defining and measuring poverty in the country, while the second updated the results in light of the next household survey.

The limited independence of local poverty measurement

This did not mean that poverty measurement was independent from globalized conceptions of poverty. Much of the ostensible ‘national independence’ of those first poverty studies (Abū Khalīl, 2005) emerged by default rather than design. The authors of the first poverty report deliberately sought to follow ‘international criteria/standards’ in calculating basic needs in Jordan, referring to unspecified ILO and World Bank studies, as well as ‘studies implemented in some developing states’ (MoSD, 1989, p. 80). As the team leader recalled, they even sought for help from the World Bank, UNDP and other agencies but did not receive a response.² Connections between the ministry and UNDP or the ILO were feeble at the time, and this search for help coincided with the heyday of the World Bank’s Structural Adjustment Policies, which had temporarily dropped the poverty agenda from its priorities. The team’s search for comparable studies on Arab or other countries, which they might be able to use as a blueprint for their own endeavour, was also largely futile. So they ended up having to teach themselves.³

Comparing the methodology and framing of the first two ‘national poverty studies’ with the World Bank’s subsequent ‘Jordan Poverty Assessment’ of 1994, it becomes clear that while the reports differ in some regards, these differences are largely rooted in expediency rather than in a differing ‘local’ perspective on poverty. A crucial commonality in all three reports (1989, 1993, and 1994), which set the tone for all future poverty measurements in Jordan, was the attempt to determine poverty rates by setting absolute (monetary) poverty lines.⁴ According to the national study’s team leader, who was aware of the diversity of possible definitions of poverty and ways to calculate it but certainly not of the full extent of all the options available at the time,⁵ this choice was a practical one that resulted from difficulties in measuring relative poverty. Yet he also assumed that using absolute poverty measurements ‘may be more appropriate for the purposes of planning and follow-up on development programs that aim to satisfy basic needs’ (Al-Saqour, 1990, p. 115). This basic methodological choice intentionally corresponded with the World Bank’s preferred method of poverty measurement. The 1994 World Bank’s ‘Poverty Assessment’ as well as subsequent World Bank-commissioned reports, in turn, relied on it because its methodology expressly came closest to what was later standardized as World Bank methodology (Shaban et al., 2001, pp. 33–35). The MoSD team and the World Bank thus never held fundamentally different positions regarding what constitutes poverty and how to determine it.

There was, however, a notable difference in terms of framing interventions to reduce poverty. The first, ‘home-grown’ reports shied away from giving policy recommendations. Nevertheless, they served as a basis for policy-makers to publicly discuss poverty as a social phenomenon – at the time in a relatively diverse way reflecting various intellectual traditions (see e.g. Amr & Al-Dwayk, 1987; Ḥassan & Al-Barghūthī, 1990). They also provided the basis for establishing the publicly-funded National Aid Fund to provide cash assistance to those identified as poor, although on the basis of status categories rather than income.⁶ In contrast, the first World Bank-commissioned ‘Poverty Assessment’ (WB, 1994) represented an attempt at streamlining poverty alleviation discourse in and for Jordan with the globalized language of structural adjustment combined with targeted poverty alleviation. It held previous social policies like government employment, generalized subsidies, price controls, and trade restrictions primarily responsible for increasing poverty, while omitting the effects of IMF fiscal austerity measures entirely (WB, 1994, p. xi). Instead, growth was promoted as a universal remedy and as a fundamental precondition for poverty alleviation. This was combined with a depiction of the social safety net philosophy as a temporary aid for those considered as poor. This discourse, largely reiterated in subsequent World Bank-commissioned reports on Jordan (see e.g. WB, 2004a, p. 6) closely resembles ‘poverty assessments’ issued for other borrowing countries (see Hanmer et al., 1999).

Poverty reporting thus shifted focus, moving away from poverty and its characteristics and toward the fallout of globalization and the policies required to encourage pro-poor growth (Abū Khalīl, 2005). Poverty line methodologies helped form these arguments, as they allowed for a decontextualized, depoliticizing perspective on the origins of poverty, and dispelled the language of universal rights and citizenship that is always latent when discussing poverty (cf. Destremau, 2001; Lautier, 2013). The 1994 Poverty Assessment omitted or dismissed alternative ways of profiling poverty, such as subjective poverty estimates, and thereby reasserted that measuring poverty through absolute monetary poverty lines was the only feasible strategy (WB, 1994, pp. 168–174). As a British consultant pointed out, this choice was not just useful from the perspective of the World Bank at the time. It also enabled successive Jordanian governments to engage with the problem of poverty without rocking the political boat,⁷ displacing questions of welfare retrenchment and the narrowing social base of the Hashemite monarchy during structural adjustment. It helped obfuscate the socio-economic exclusion of rural East Bank constituencies, who were now increasingly addressed through targeted projects instead of more universal forms of social provisioning and public sector employment (see Baylouny, 2008).

Poverty indicators, however, remained contested for years. A poverty analyst who used to work at the Jordanian Ministry of Planning and International Cooperation (MoPIC) recounts the trajectory of poverty studies in the 1990s, i.e. after the nationally-led 1989 study: ‘Then we started having more reports, consultancies, and a number of institutions started to be interested, like the UNDP, the ESCWA, the World Bank. You know, everybody wanted to have his own approach and methodology, and we have many reports in the early 1990s.’⁸ Some of these methodologies simply put forward competing poverty lines and rates. Others, most prominently the Jordan Human Development Report (JHDR) commissioned by UNDP, broadened their scope to link discussions on poverty with issues like social exclusion, inequality and government policy (MoPIC/UNDP/JOHUD, 2004; but see De Bel-Air, 2011).

The existence of various international organizations or donors with different and evolving standards for poverty measurement carved out some space for different representations and measurements of poverty in Jordan. While the World Bank methodology remained dominant, this allowed for alliances between Jordanian poverty experts, international consultants and different international organizations who pushed for other ways of understanding poverty and specific types of intervention. This latter group sought to validate their perspective by referring to different international standards, such as multidimensional understandings of poverty and the Participatory Rural Appraisal methods that were heavily used in the 2004 JHDR. Yet these critical voices never became dominant (see Lenner, 2014). As one analyst involved in the JHDR noted, ‘we keep on […] saying that we need to influence decision-makers, but it’s not working to be honest.'⁹

In sum, tracing the history of poverty measurement in Jordan shows that much like previous expansions of statistical capacity, the establishment of poverty measurement in Jordan was driven by Jordanian policy-makers and statisticians’ efforts to render legible a phenomenon that came on their radar during a period of crisis, in order to devise ways of tackling it. While initially relatively open in terms of narrative, the use of monetary poverty lines facilitated the process of making poverty legible and governable in line with the perspective of World Bank experts, who sought to establish a clear narrative of poverty alleviation as part of structural adjustment policies. This narrative helped Jordanian rulers obfuscate the gradual erosion of previous forms of welfare provision, and the regime’s changing social support base.

Ambiguity, transparency and opacity in setting ‘acceptable’ poverty lines

The choice of one type of socio-economic indicator over another has profound political implications, yet so do specific methodological choices underlying the composition of each of these indicators. This becomes obvious when exploring the creation and calibration of the globally dominant indicator of poverty until today: the (absolute) national poverty line and, derived from that, the headcount poverty rate.¹⁰As a money-metric indicator, it tends to be seen as the most objective form of measuring poverty by international and national policy-makers. This is why, despite the now near universal agreement among experts that poverty is multidimensional, and despite multidimensional indicators being readily available in Jordan and globally, it continues to function as the primary yardstick for efforts to reduce poverty (see Sumner, 2007). Yet despite this semblance of objectivity, the creation of money-metric poverty lines and rates is a highly contentious process. As Fischer highlights, ‘official poverty lines need to be understood first and foremost as the outcome of complex processes of political contestation’ (2018, pp. 103–104). In this section, I trace how this process combines transparency and opacity, both of which are central to reproducing the Jordanian state as an actor principally capable of tackling poverty.

The politics of methodological ambiguities

The poverty line in Jordan is derived from data collected by the Household Expenditure and Income Surveys (HEIS). Uncertainties encountered during the administration of the HEIS have existed since the first surveys were administered in the 1980s, alongside doubts about the representativeness of their findings, and their international comparability among some donors and data producers up to today.¹¹ These doubts have at times been used as a pretext for stronger donor intervention into processes of data collection and aggregation. But they have not stopped either donor or governmental agencies from relying on the data the HEIS has produced. They have used them to create global rankings, seek and dispense aid, and legitimize operations in the name of poverty alleviation. And in doing so, they have allowed questions around the validity of the data to disappear (see also Bonnecasse, 2011).

Secondly, the process of constructing poverty lines and rates on the basis of those raw data is fraught with ambiguities. The same raw data can be aggregated and interpreted very differently, depending on precisely how the poverty line is constructed. This ambiguity has made it possible to accommodate a range of considerations in changing political contexts. As the team leader of the first poverty study in the country noted: ‘We always have a base, a political guideline, how to write your statistics and research – they can give you high poverty; others say no, low trend’.¹² To illustrate, he described how the World Bank consultants, together with MoPIC and the Jordanian Department of Statistics (DoS), repeatedly ‘brought down poverty’ in the early 2000s. This was just after King Abdallah’s ascension to the throne, when ruling elites as well as IFIs were seeking to portray Jordan as a middle-income country undergoing sweeping economic reforms, and in need of foreign investment to do so. This was in marked contrast to his negotiations with the head of MoPIC a decade earlier, in the early days of structural adjustment, who asked him to raise the poverty rate in order to make a better claim for World Bank assistance during a fallout between Amman and Washington in the course of the 1990 Gulf War.¹³

The specific choice of methodology has had a decisive bearing on the figures and the political messages resulting from them. To illustrate this, it is revealing to look at two studies that both used data from the 1997 and 1992 HEIS to determine the poverty line(s) and the respective poverty indicators (see Table 1): The first study, which was released in 1999 by the Ministry of Social Development (MoSD) in cooperation with Britain’s DFID and UNDP, set the per capita annual poverty line at JD 468. It concluded using this figure that the headcount poverty incidence in 1997 was 33% (compared to 27% in 1992) (see MoSD et al., 1999). This study was never officially published and is not available in the public domain. As such, it has largely fallen into oblivion. The second study (Shaban et al., 2001), which was produced by consultants from the World Bank in cooperation with Jordanian experts, set the poverty line at JD 313.5 and the headcount poverty incidence at 11.7% (compared to 14.4% in 1992). This study was officially adopted and published. Besides the major variation in the poverty rate, the first study showed that the poverty rate rose significantly between 1992 and 1997, whereas the second study showed that it declined.

Table 1. Competing poverty indicators based on HEIS 1992 and 1997.

Study	HEIS year	Calculation of food poverty line	Required calorie intake	Absolute monthly food poverty line	Absolute monthly poverty line	Absolute poverty rate	Change in poverty rate 1992–1997
MoSD et al. (1999)	1997	Actual food pattern method	2440	JD 18.6	JD 39	33%	+6.0%
	1992			JD 12.57	JD 28.16	27%
Shaban et al., 2001	1997	Least cost for recommended food basket	2224	N.A.	JD 26.13	11.7%	−2.71%
	1992			N.A.	JD 21.75	14.42%

Source: Shaban et al. (2001, p. 36, tb. AI.1); MoSD et al. (1999, 9–10, Tb. 2).

Those major differences are a result of the different methods used to determine the food poverty line, which forms the basis for calculating the absolute poverty line. The World Bank study is based on calculating the least cost for a recommended food basket to meet nutritional requirements.¹⁴ In contrast, the MoSD et al. study calculates actual food expenditures for those consuming no more than the recommended calorie intake for Jordan, who are thereby declared poor (1999, pp. 5–6). These expenditures are higher than the least possible cost, as many studies have found that the poor tend to pay more for goods and services than the rich even in the same locale (cf. Hanmer et al., 1999, p. 802). Furthermore, the MoSD et al. study determines the food basket differently. It sets a higher required calorie intake level, discounts a fraction of the food consumed to loss and spoilage, and keeps expenditures such as recreation and personal care in the list of basic needs, all of which contribute to raising the poverty line (MoSD et al., 1999, pp. 15–17; Shaban et al., 2001, p. 35).

The methodological differences, and their results, point to different agendas regarding the representation of poverty. Showing the ‘right’ trend in poverty levels is an important part of this endeavour. It functions like a commentary on the socio-economic situation, as well as the specific policies being pursued at the time. The two studies outlined above cover the decisive first years of structural adjustment in Jordan. Poverty decreased with the methodology chosen by the World Bank, whereas it increased in the case of the alternative study. Demonstrating the positive effect of the World Bank’s policies arguably constituted a powerful motive for choosing a poverty line that would reveal a downward trend, both for the World Bank and the Jordanian government (see also Harrigan et al., 2006a). The MoSD et al. study lacked this motivation. The MoSD was losing political clout in this period and DFID was not as relevant a donor as the World Bank, so its results could easily – and conveniently – be forgotten. Interestingly, this did not stop the World Bank report from denouncing the methodology used in the MoSD et al. report, claiming for example that its food basket was based on ‘arbitrary assumptions’ (e.g. Shaban et al., 2001, p. 35). Whether or not this convinced anyone, it still shows one actor-network seeking to destabilize another by highlighting the ambiguities in the creation of its indicator, thereby undermining its seeming objectivity and the political commentary it suggests.

There was no officially agreed on poverty line and rate throughout the 1990s and early 2000s. Different actor-networks, composed of local and international statisticians, Jordanian governmental institutions, international organizations, and different donors, calculated their own poverty measurements. This occasionally produced remarkably different representations of poverty. This changed in 2003, when Jordan’s Department of Statistics (DoS) legally became the only agency allowed to collect and publish statistical data (or authorize another to do so). The years 2002/3 were characterized by a deliberate effort to refocus public attention on domestic, particularly socio-economic issues after the second intifada in Palestine and the start of the war in Iraq, but also after sustained protests around political repression and socio-economic frustrations in the south of Jordan (Harrigan et al., 2006a, p. 286). As a socio-economic researcher convincingly argued, this must have triggered a desire to tighten control over socio-economic data production and publication, so as to be better able to make authoritative, seemingly objective statements about poverty.¹⁵

Yet this did not mean that poverty indicators were now produced by Jordanian statisticians only. There was no separate poverty division at the DoS until 2009, which meant that a World Bank expert, in coordination with MoPIC’s policies and strategies department, decided on the methodology to be used and calculated the poverty line and rate. After 2009 DoS staff took over the calculation process, yet MoPIC continued to decide on the methodology and narrative used for each poverty status report, and a World Bank consultant continued to pitch in at crucial points.¹⁶ What consolidating authority under DoS did mean was that there were no longer competing monetary poverty indicators, and that data production and use was removed not just from local social relationships, as Mitchell (2002, p. 100) suggests, but also from alternative expertise on how monetary or more multidimensional indicators could be tabulated, and to what end.¹⁷

Maintaining acceptable poverty rates over time

Despite the tightening of numerical control, ambiguities regarding data and methodologies continued to be negotiated between (a few) statistical experts, government agencies and donors. A number of interview partners insinuated that this led to long delays in the release of poverty figures, or to blocking their publication altogether.¹⁸While insights into the details of these negotiations are anecdotal, their results have been remarkably consistent over time. Since 2001, published poverty indicators have remained within a range that is ‘not too high, not too low’. I take this phrasing from the World Bank’s 2004 ‘Jordan poverty assessment’, where the reasoning for determining the absolute poverty line is spelled out as follows:

If the poverty line is too high, then the government would not be able to fulfil its commitment to maintain the minimum standard of living of its population. The poverty line should not be so low that everyone in the society is non-poor. In such a situation, the government may not be motivated enough to raise the standard of living of those who are unable meet the absolute basic needs. (WB, 2004b, p. 10)

This reasoning is remarkably candid on the fact that setting an absolute poverty line, supposed to objectively define the minimum living standard in a country, is subject to political considerations (more on those below). The resulting poverty rates – consistently between 11.7% and 15.7% – have equally remained within a seemingly acceptable range.

In creating those indicators, transparency and opacity have both played important roles. Each poverty report analysed goes to great lengths to lay open the methodology adopted and justify specific methodological choices. This transparency bolsters the indicators’ credibility. Even highly critical poverty experts in Jordan, who found the results hard to stomach given their own experiences, conceded that the methodology for calculating the 2002 poverty line was sound, for example. They did not believe the numbers had been fabricated.¹⁹ So at least within the community of professionals, the practice of making methodologies of calculation transparent can, and in this case did, have the effect of insulating the resulting figures from fundamental critique.

At the same time, these ‘acceptable’ poverty indicators have remained opaque, particularly when considering their development over time. This is particularly due to several recalculations of the poverty line. These have, in effect, kept the poverty rate ‘not too high, not too low’ and demonstrated favourable poverty trends, yet also made it difficult to compare indicators over time. In 2004, the World Bank and DoS recalculated the 1997 poverty line discussed above on the basis of a new methodology. The report justified this on the grounds of (international) comparability and compliance with up-to-date World Bank methodologies. Instead of JD 313.5, the poverty line in 1997 was now JD 366 per person per year. The report concluded that, as a result, the poverty rate in 1997 was actually 21.3%, rather than the 11.7% it had originally calculated (see Table 2). It then went on to compare these recalculated figures with new data for 2002, which revealed a poverty incidence of 14.2%. This led to the conclusion that ‘poverty reduced in Jordan significantly’ between 1997 and 2002 – the third phase of Structural Adjustment (WB, 2004b, p. 28). This ‘success’ would not have been possible if they had compared the new data directly with their previous report.

Table 2. Recalculations of official poverty rates.

HEIS	Shaban et al. (2001)	WB (2004b)	WB (2009)	DoS (2010)	DoS (2012)	Bibars (2019)
1997	11.7%	21.3%
2002		14.2%	20%
2006			13%	13%
2008				13.3%	(10.5%)
2010					14.4%	?
2017/8						15.7%

Note: Rates in bold represent poverty rates as they were calculated in the original report. Rates in italics represent the previous HEIS’s findings (the rate to their left) recalculated based on the new methodology/ base year.

As Table 2 illustrates, the recalculation of poverty lines and rates has been a rather constant feature of poverty measurements ever since. Poverty lines were recalculated again with the 2006 HEIS. The resulting report, which reset the fixed year to 2006, showed a poverty rate of 13% for 2006. The 2002 poverty line was recalculated to reflect the change, and the readers are informed, in a footnote, that the recalculation yielded an estimated poverty rate of 20% – rather than 14.2% for 2002 (WB, 2009, p. 8). Again, this recalculation allowed for the conclusion that that ‘poverty rates have fallen between 2002 and 2006’ (WB, 2009, iii). The same methodology and same fixed year were used in 2008, and showed a very slight increase of poverty rates between 2006 and 2008. But the analysis of the 2010 HEIS changed the methodology yet again. It switched from using the lowest 20% to the lowest 30% of the population as reference group for calculating consumption (cf. DoS, 2012, p. 7), arriving at an increase of poverty levels that would have likely been much more significant had the methodology remained the same. According to a statistician involved, this methodological shift was suggested by the World Bank consultant and then approved by MoPIC. This brought the method more in line with that used in other middle-income countries (see note 15; Jolliffe & Serajuddin, 2018); yet it – again – obstructed comparisons over time and painted a more favourable picture than maintaining the same methodology would have. This time, the recalculated indicator for 2008, which showed a lower poverty rate of 10.5% (compared to 13.3%) in a draft version, was not even mentioned in the final report, thereby avoiding the comparison altogether. This came after delaying the publication for two years, reportedly due to concerns that the figures suggested the government had not succeeded in reducing poverty despite its heavy engagement on that front.²⁰

In the last decade, opacity has further increased. The 2014/15 HEIS, eagerly awaited by the donor community because of what it would reveal about the effects of the Syrian crisis, was retrospectively declared invalid by the government because of concerns about the adequate representation of Syrian refugees in the sample.²¹ DoS conducted a new survey in 2017/8, and on its basis a new poverty rate of 15.7% was announced in 2019. The head of the DoS stated that the methodology had changed for this calculation, but how it changed was never revealed (ARDD, 2019; Bibars, 2019). Withholding such information has been made easier by the changing interests of donors and IOs. Instead of poverty, their focus in recent years has been on social protection and vulnerability in conjunction with Syrian displacement (see e.g. HKJ/UNICEF, 2019; UNICEF, 2020). Yet poverty rates are still considered crucial for representation to Jordanian publics, and the figures announced in 2019 have been criticized as unrealistic given the severe economic crisis that Jordan has undergone in recent years, and despite the supposed ‘win-win’ solutions of the Syrian refugee response (Bibars, 2019; Lenner & Turner, 2019). In the domestic realm methodological choices for setting the poverty rate remain highly relevant.

The state effect of transparency and opacity

Frequent changes of methodology, in combination with restricted data access, in effect mean that you can only ever compare one report with the report immediately preceding it. The whole series is never included in the recalculations.²² Withholding methodological information entirely makes comparison even harder. Whether the result of a deliberate decision or rather, as DoS staff implied, a combination of negligence by policy-makers, who only want current numbers, and the relative inexperience and limited clout of the staff, this makes it possible for current results to always remain within the bounds of the ‘not too high, not too low’ range. Those figures make poverty seem manageable and managed. They contribute to the effect of the Jordanian state as a free-standing entity that is capable of holding socio-economic inequalities in check and taking care of its citizens, while being in need but also worthy of continued donor support.

The way in which these figures are calculated creates a degree of transparency and credibility, yet also makes it impossible to identify long-term poverty trends and relate these to specific policies. There is a section in every major poverty report that details how various government interventions have reduced poverty rates since the preceding report, suggesting that without these interventions the headline numbers would have been considerably higher (e.g. DoS, 2010, pp. 75–82; DoS, 2012, pp. 94–99; Shaban et al., 2001, pp. 15–21). These snapshots effectively signal governmental action, but the lack of comparability over time makes it impossible to gauge the longer-term effects of these policies. This opacity disorients sustained critique.

Notwithstanding their intransparency, published poverty indicators continue to render poverty legible and serve as a basis for governmental action. Since 2004, their breakdown by subdistrict has fed strongly into the so-called ‘poverty pockets programme’. This flagship anti-poverty programme ran until at least 2013 with the help of substantial government funding and donor support (see Lenner, 2013). It did not decisively contribute to lowering absolute poverty levels across those poverty pockets (according to the available indicators), but it did reinforce the representation of a state that takes a systematic approach to poverty. At the same time, the unavailability of a time series served to deflect questions about why rates changed in specific poverty pockets.²³ In short, the patchy data allowed governing actors to effectively shirk the question of responsibility.

All of this has not changed the widespread incredulity of the broader population, whose perception of constantly rising poverty levels contradicts official poverty figures. Explicit remarks to this end in official poverty reports (e.g. WB, 2009), media commentaries following the release of new poverty status reports, as well as slogans during demonstrations attest to this widespread scepticism (see Lenner, 2014, pp. 187–182). It also has not made the role of international financial institutions fade into the background. Indeed, IMF-imposed economic policies impoverishing the country’s inhabitants have been the subject of major countrywide protests in recent years. However, even those cross-class alliances continue to portray the state as responsible for its inhabitants’ wellbeing and capable of changing the current situation with the right measures (see Ababneh, 2018). As Martínez argues, this form of ‘responsibility talk’ (2022, p. 151) also reproduces the state: ‘Citizens may relate to the state through compromises, complicity, cynicism or sincere appeal, but they do not cease to reproduce the state as the great enframer of their lives’ (2022, p. 167). Poverty indicators are certainly not the only governmental technique contributing to this effect; myriad daily interactions reproduce the state as a coherent entity regulating socio-economic life. They do, however, help maintain the sense that there is a state responsible for socio-economic policy-making for the benefit of its citizens, while at the same time insulating it from fundamental critique through obfuscation.

Conclusion

Poverty indicators for Jordan have been marked by a mixture of opacity and transparency since the very beginning of their tabulation. This has allowed their producers to accommodate a variety of agendas and agencies as well as diverse forms of political messaging. On the whole, the use of monetary poverty indicators, whether calculated by Jordanians, World Bank consultants, or both, has facilitated a depoliticized understanding of poverty as an individual problem that can be solved through targeted measures accompanying structural adjustment. Competing interpretations put forward by alternative networks, and more politicized multidimensional forms of poverty measurement, have never managed to become dominant.

Opacity as a governing technique has increased since data collection and analysis were centralized under the DoS. While this has not firmly localized data production, it has widened the scope for involved actors to render poverty in Jordan opaque. That said, transparency continues to play a crucial role in the creation and calibration of these indicators. Poverty rates resulting from this mix of transparency and opacity have consistently been ‘not too high, and not too low’ and served as grounds for continued governmental action. They have contributed to the effect of a state that is, in principle, capable of and responsible for tackling socio-economic inequalities, while shielding it from closer scrutiny. Thereby, they have helped buttress authoritarian power during a period marked by crisis, structural adjustment and welfare retrenchment.

The Jordanian example highlights the need to more explicitly include questions of methodology in the debate on governing through indicators, and demonstrates the utility of combining discussions on indicators with scholarship on transparency, ambiguity and opacity. From this vantage point, we can better understand the mechanics producing globalized socio-economic indicators and the effects they have. This case study also provides a lens for analysing a crucial dynamic through which the state effect emerges in the Middle East, and in other parts of the world. Acknowledging how transparency and opacity combine to produce acceptable socio-economic indicators further contributes to de-exceptionalizing the Middle East in discussions about the globalized politics of development.

Acknowledgements

I would like to thank my interlocutors for their time and patience, as well as five anonymous reviewers, the special issue editors and Cameron Thibos for their valuable comments and suggestions.

Disclosure statement

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

Additional information

Funding

This work was supported by DUNIA BEAM - Erasmus Mundus; DUNI1301743]; EU Sixth Framework Programme, FP6 - CITIZENS, RAMSES II [grant number 513366]; German Academic Exchange Service (DAAD) [grant number (Z) 55414388].

Notes on contributors

Katharina Lenner

Katharina Lenner is a lecturer in Social and Policy Sciences at the University of Bath. Her work focuses on the governance of socio-economic development, labour market participation and (forced) migration in the Middle East, and in Europe.

Notes

1 The Jordanian Five-Year-Plan for 1981–1985, for example, does not mention the word ’poverty’, and is largely focused on economic growth and investment. It does pick up on the discourse of basic needs though, and stresses the need for a more equitable distribution of incomes (NPC, 1980).

2 Interview with former MoSD Secretary General, March 2010, and Jordanian journalist and poverty researcher, March 2010.

3 Interview with former MoSD Secretary General, March 2010, and Jordanian journalist and poverty researcher, March 2010.

4 Absolute poverty lines determine a fixed sum needed to satisfy basic needs, which is independent from the wealth of the rest of society. They usually comprise (a) an abject poverty line, which is equivalent to the average cost to satisfy basic food needs; and (b) a poverty line for basic non-food needs. Together they form the absolute poverty line. In comparison, relative poverty lines determine poverty levels in relation to the wealth of the rest of society (e.g. Hanmer et al., 1999, pp. 797–800; Ravallion, 2010).

5 For an in-depth perspective on the development of social statistics at the UN, see Ward (2004).

6 Interview with former MoSD Secretary General, March 2010

7 Interview with consultant for JOHUD, January 2008.

8 Interview with poverty analyst, formerly at MoPIC, November 2009.

9 Interview with consultant for JOHUD, November 2006.

10 The headcount poverty rate is synonymous with the ‘poverty incidence’ when the latter is calculated with the individual as basic unit of analysis.

11 E.g. interview with Social Statistics and Poverty Studies Consultant for DoS, April 2012; see also WB (1994, pp. 172–174), Harrigan et al. (2006a). A recent study using the data for the 2010 HEIS concluded that the quarterly survey intervals end up underestimating poverty rates by more than 25% in comparison with the globally much more common annual survey intervals (Jolliffe & Serajuddin, 2018).

12 Interview with former MoSD Secretary General, March 2010.

13 Interview with former MoSD Secretary General, March 2010

14 This study did not do all the respective calculations itself but basically adopted the methodology of the ‘national’ 1987 and 1993 studies, and updated its poverty line based on current prices. Even according to World Bank standards, there is no single way to determine the food poverty line, yet even so the method adopted in Jordan until 2012 is unusual when compared to those of other countries under WB guidance. It uses the food expenditures of the poorest quintile of the population as a reference point for calculating the food poverty line, rather than what it would cost to satisfy food needs as a non-impoverished person. This, in effect, considerably lowers the food poverty line. The overwhelming majority of middle-income countries uses a broader reference group in order to calculate this line (see WB, 2004b, pp. 20–22, table 1.10; El-Rayyes, 2007).

15 Interview with associated researcher, IFPO, October 2012.

16 Interviews and informal conversations with poverty analysts at the DoS, March 2011 and February 2015, and at MoPIC’s social studies division, March 2015.

17 Interviews with poverty analyst at the Coordination Commission for Social Solidarity, Apr. 2012; Consultant with JOHUD, January 2008.

18 Interviews with associated researcher, IFPO, December 2009 and poverty analyst at the DoS, March 2011

19 Interview with senior managers of ZENID and JOHUD, November 2006.

20 Interview with former staff of MoPIC’s Economic and Social Productivity Program (ESPP) unit, May 2014.

21 Interview with policies and strategies department, MoPIC, June 2016.

22 Interviews and informal conversations at the DoS, March 2011 & February 2015.

23 Interviews and informal conversations at the DoS, March 2011 & February 2015.