The Data War Comes Home – Developing Economics

The Trump administration’s ongoing attempts at manipulating US government economic data echoes controversies that have existed in the realm of development data for decades. These controversies highlight the unavoidable, intrinsically political nature of measuring social phenomena with economic statistics, and the role of economists in legitimizing (or not) such measures.

In early August, the employment report from the Bureau of Labor Statistics (BLS) showed weaker than expected job growth for July and announced downward revisions (fewer new jobs than previously reported) for the two months prior. Trump responded by calling the accuracy of the report into question, and firing the head of the BLS, a career civil servant. This move provoked a round of criticism from other civil servants, economists, and experts on democracy, which only intensified when Trump initially nominated as the replacement E.J. Antoni, the chief economist for the right-wing think tank The Heritage Foundation. Antoni had been an outspoken critic of the BLS reports, even suggesting the possibility of ceasing to release monthly jobs data altogether, and was widely perceived by critics as both highly partisan and underqualified for the position. There is widespread concern that the jobs report will become less reliable, even leading to the current staff at the BLS publicly pleading with the public to still trust their numbers, for now. The BLS is also responsible for producing the Consumer Price Index (CPI), the central measure of inflation produced by the US government. Inflation’s centrality in recent politics, including promises from Trump to bring down prices on “day one,” have led to concerns that this measure could also be affected by political manipulation. There would be important real-world and policy impacts of a degraded CPI measure,  which affects tax brackets, the value of some treasury bonds, and social security and other social insurance payments.

The “War on Government Statistics”

The BLS firing is part of a larger pattern of manipulation and misuse of government-produced data. During the first Trump administration, the council of economic advisors used a change in the CPI to lower the poverty line, then claimed the end of poverty justified reductions to the social safety net. More recently, members of the administration and some of their supporters (including Antoni) have pointed to evidence of a massive increase in “native-born” employment in the BLS monthly reports. This apparent increase, according to the Trump faction, is reason to ignore an overall weakening labor market since jobs are going to the “right” people. The claims of the massive increase in employment for native-born workers, however, are implausible on their face, and not supported by other data such as the unemployment rate. A closer look at the data reveals that the supposed large increase in jobs for native-born Americans is actually a statistical artifact related to how the relevant surveys are weighted and updated annually, and perhaps to increased non-response by non-citizens.

Various types of data and data collection practices inconvenient to the current administration’s political objectives have also begun to disappear altogether, including data related to race and ethnicity of federal employees, the collection of data related to sexual orientation and gender identity for disease surveillance, and climate change reports from NASA’s website. Government statistical offices have also been degraded through budget cuts and staff reductions. Trump dismissed technical advisory boards to the BLS, the Commerce Department’s Bureau of Economic Analysis and at the Census Bureau. The sum total of these actions was described by Sahm as a “War on Government Statistics.”

The “Authoritarian Playbook”

Politically motivated manipulation of economic data has often been portrayed as mainly the domain of authoritarian, populist, or otherwise politically disfavored countries. Many critics of Trump’s BLS moves have couched their criticisms in such terms, including some of the elite of the economics profession. Former treasury secretary and Federal Reserve chair Janet Yellen described the move as “the kind of thing you would only expect to see in a banana republic,” and former treasury secretary and Chief Economist of the World Bank Larry Summers remarked that it was “what happens in authoritarian countries, not democratic ones.”

Recent history does offer some examples of this correlation. Gladstein tells the story of such regimes that influenced, or outright faked, statistics for a variety of political reasons. In 2022, Recep Tayyip Erdoğan of Turkey fired his government’s statistics chief after a report showing high inflation, as did Argentina’s Cristina Fernández de Kirchner, leading to  a widespread loss of confidence in the accuracy of Argentina’s economic data, and  censure from the IMF. China has historically been accused of a lack of transparency in GDP calculations, and many studies have indicated that local jurisdictions have overrepresented their economic output for political purposes.  (Though this type of direct manipulation has likely now been minimized). Beyond such individual anecdotes, there have been some attempts to find more systematic evidence of this phenomenon. Doces and Magee  find that governments coded as authoritarian were systematically more likely to exaggerate their economic growth in submissions to international agencies.

Development Statistics and Power

The focus on only governments deemed authoritarian, however, misses the fact that more “respectable” actors, such as multi-lateral development institutions and wealthy world governments, have also taken part in the politicization, misrepresentation, and manipulation of economic data. How social phenomena are defined, and measured, and who gets to decide the definitions and measurements, are inherently political questions. Numerical measures can be presented as unambiguous, objective empirical fact, but often have values, interpretative choices, or theories of the world embedded within. The power relationship intrinsic to the collecting and dissemination of data and the use of technical measurements is especially clear in the domain of international development data, where these choices are often made by experts far removed from the individuals being measured, in a context in which unequal power relationships prevail.

Multi-lateral institutions such as the World Bank and United Nations agencies play a role in legitimizing false or manipulated data from member states. Jerven explains how discrepancies between national statistics offices and international institutions such as the World Bank were often settled by political negotiations. National governments may then use this process to “launder” their false or misleading statistics; the government provides data to a multilateral institution such as the UN or World Bank, these institutions use that data to produce, say, the HDI, or reports on progress towards the SDGs, or a database on economic growth rates, and the governments in turn can point to these more “credible” sources as proof of their own success.

Multilateral development organizations have also been accused of direct manipulation in accordance with the political agendas of their member states. A change in how the Bank calculated China’s GDP in 1999, which subsequently allowed China to borrow more cheaply due to the Bank’s classification system, has been described by Wade as “unexplainable by anything other than political motives.” Wade also tells the story of how the Bank’s World Development Report in 2000 was altered to lessen its focus on inequality after political pressure. At The World Bank research department, Broad contends that researchers were discouraged from publishing data that contradicted the dominant political economic paradigm, and that manipulation of data to similar ends may have also occurred. In 2018, the Bank’s “doing business” ranking, which purports to calculate a numeric score indicating how conducive a country’s regulatory environments are for business, came under scrutiny. The method of calculating the index was repeatedly changed in a short period, and those changes resulted in a sharp drop for Chile. Chile was governed at that time by the socialist Michelle Bachelet, yet the massive changes in ranking were almost entirely based on the change in methods, and not any actual changes in Chile’s regulatory environment.

Aside from direct manipulation, the choices that multilateral organizations make in terms of what data to collect and measure, and what type of measurements to choose for indicators and targets, are also affected by politics and ideology. The choice of measurement is a choice of the meaning of the concept, and sets standards by which progress is measured and policy outcomes are assessed. The correct way to measure global poverty, for instance, has been the subject of debate for decades, due for example to questions of whether poverty should mainly be considered relative to national circumstances or only in a global context, or to questions about how to define poverty, and who gets to define it. The way that the Bank was calculating global poverty rates for decades was shown to be highly dependent on various measurement choices, and on definitions of what constitutes poverty that were detached from real human needs. Cobham also documents how individuals from disadvantaged groups are often underrepresented in economic data globally, which both leads to an understatement of inequality, and also exacerbates such inequality when leading to political underrepresentation or under-provision of services. In previous work, I’ve shown how even the seemingly straightforward question “is global inequality falling” is highly dependent on the measurement choice, which is in turn influenced by values, moral judgements, and political imperatives. The most commonly used measures of inequality are more likely to show falling or lower global inequality when compared to other, equally plausible options.

The choices of indicators used by the international development agencies, such as for the Millennium Development Goals, are influenced by politics, and also have real world effects. For the Sustainable Development goal related to reduced inequality, for example, the UN chose a comparison between the growth rate of the bottom 40% and the average growth rate of the population. Many critics have argued that this choice of measurement is more politically palatable to some UN members, as compared to a measure that includes the incomes of top earners, or one that takes into account levels of inequality rather than just rates of change.

Data Manipulation in the “Core”

Wealthy democracies, and their institutions, have also not previously been innocent of such data manipulation and politically motivated measurement choices. In one example, the Obama administration proposed a switch to using an alternative method of calculating inflation for the purposes of determining “cost of living adjustments” to social security payments. Proponents of this change claimed, dubiously, that it would be more accurate. However, since this change would result in smaller annual increases in the payments, critics viewed this change as a “backdoor” method of achieving otherwise politically difficult cuts to social security.

Greece, with help from Wall Street giant Goldman Sachs, famously used accounting manipulations to report substantially lower deficits and debt in order to (appear to) meet the “Maastricht” criteria for joining the European Union. Italy had already used similar accounting tactics to lower their deficits for the Maastricht goals, this time with the help of JP Morgan, and Goldman again. Some informed observers argue that officials in Brussels and other EU countries willingly went along with the deception in order to achieve the political goal of Italy’s EU entry (See e.g. Varoufakis pgs. 131-2). In the years leading to the global financial crisis, such dishonest reporting continued in both countries, with the continued assistance of both banks, so as to avoid penalties for running afoul of EU legally required debt and deficit limits. Post-crash, when Greece’s head of national statistics updated the figures to reveal a substantially larger deficit, he was accused of lying and criminally charged.

Technocratic Development, Economists as Governors

The importance of data plays a central role in development economics partially due to the long term ideological and political turn away from a more comprehensive idea of development and towards a technocratic effort to reduce poverty and induce improvement on specifically defined goals and indicators. This dependence on statistics and data has in turn helped give economists special status in development debates. The credibility of economic or development measurements will relate to their endorsement by prominent academics or expert communities. If indicators are tools of governance, write Davis, Kingsbury, and Merry , then those who influence the form of said indicators should be “counted among the governors.”

Economists similarly play an outsized role in using research to shape how social problems are understood in the US, a privileged status at times jealously guarded by elite economists. There were no shortage of legitimate criticisms of  Trump’s now withdrawn appointment of Antoni at the BLS, yet some in the profession chose to focus on his lack of proper credentials as an elite economist, as indicated by his alma mater and lack of publication history. When Biden appointed an economic advisor deemed too politically left in some circles, she faced similar criticisms about having the “wrong” type of credentials (working for policy advocacy think tanks instead of publishing in elite journals). Such criticisms ring especially hollow when remembering that sometimes economists at even our most elite institutions forget how to use Microsoft excel properly, and at other times may engage in conspiracy to defraud the US government. A more compelling criticism might have acknowledged that the issue at hand is inherently political, and focused more on Antoni’s documented history of dishonesty in pursuit of his political agenda, rather than credentialist gatekeeping.

Conclusion

Trump’s attempts at influencing the economic data produced by the US government are clearly authoritarian in intent, and a degradation of current practices in the US. Condemning these actions on these terms is useful and necessary. However, a clearer understanding of the issue should acknowledge such actions as a continuation of practices that have been tolerated or ignored when politically useful or when committed by actors seen as more respectable. This pattern is especially clear in the area of development related data, which is often marked by unequal power relationships and a focus on technocratic goals with insufficient acknowledgment of the underlying politics.

Joshua Greenstein is Associate Professor of Economics at Hobart and William Smith Colleges.