Do geopolitical interests affect how financial markets react to IMF programs?

Abstract

We study the effect of geopolitics on short-term financial market reactions to IMF program approvals. If IMF programs are influenced by geopolitics, they may be less successful in stabilizing the economy. This could lead financial market participants to sell the country’s assets and thus reduce the catalytic effect of IMF programs. Using a monthly panel data set for about 100 IMF members covering 1993-2019, we find that if geopolitics are involved, the approval of a new IMF program increases risk aversion of financial market participants. To measure geopolitical interest, we focus on program approvals for temporary members of the United Nations Security Council (UNSC). We find that temporary UNSC members receiving an IMF program face higher bond and bill yields, depreciating exchange rates, and weaker stock market developments. This is consistent with investors reducing exposure to the country’s financial assets. Such a negative investor reaction is not observed for IMF program approvals for non-UNSC temporary members.

Keywords:

• IMF
• lending facilities
• United Nations Security Council
• catalytic effect

Introduction

We study the effect of geopolitics on the short-term financial market reactions to IMF program approvals. There is considerable anecdotal evidence and academic research on how geopolitics influence IMF lending. This is particularly true for geopolitical interests of the US, the largest IMF shareholder based on voting rights. A stark example is Pakistan.¹ According to a 2002 report of the IMF Independent Evaluation Office (IEO), US geopolitical interests strongly influenced repeated IMF lending to Pakistan. In interviews conducted by the IEO among IMF staff and Pakistani authorities, the US were perceived as supporting IMF lending for Pakistan irrespective of implementation of the program’s reforms or success. The report further states that ‘the unrealistic macroeconomic assumptions as well as the pretense of toughness were merely a way of face-saving to justify continued lending to Pakistan’ (Independent Evaluation Office of the International Monetary Fund [IEO], 2002, p. 131). Indeed, financial market participants reacted negatively to a new IMF program in 2013, a time when Pakistan was particularly exposed to US geopolitical interest. There are however also cases that do not fit this pattern, such as the IMF program for Peru in 2007, for which, despite US geopolitical exposure, no meaningful market reaction is observed.

A second, telling example of geopolitical influence in IMF lending is the US meddling in the programs of 1987 and 1991 for Egypt, which is analyzed by Momani (2004) in detail. Documents between the US Embassy in Cairo and the State Department obtained through the Freedom of Information Act show that the United States has been actively pursuing negotiations between the IMF and Egypt aimed at reducing program conditionality. This ran directly against the technocratic interests of IMF staff, who feared that such lenient programs would not achieve the economic reforms the country needed in their view. The reason for US meddling in the IMF programs was to avoid destabilization of the Egyptian government at the time because of unpopular reforms. The US considered the government an ally and essential to its regional geopolitical interests, especially regarding the First Persian Gulf War.

Why does it matter whether geopolitical interests affected the financial market reaction to IMF programs? If IMF lending decisions are politically driven, IMF programs may be less successful in stabilizing the country in question. The reason is that geopolitical decisions could undermine the signaling, or catalytic, effect of IMF lending to a country for financial markets. According to the IMF, the catalytic effect is a major goal of its lending, as ‘IMF programs can help unlock other financing, acting as a catalyst for other lenders. This is because the program can serve as a signal that the country has adopted sound policies, reinforcing policy credibility and increasing investors’ confidence’ (International Monetary Fund [IMF], 2020b). As there is considerable evidence for geopolitics often affecting IMF lending itself,² we analyze whether such interests also affect the impact of IMF lending on financial markets.

To test whether geopolitical interests affect financial market reactions to IMF programs, we use a monthly panel data set from January 1993 to December 2019 for about 100 IMF member countries, depending on the sample. We only include those countries that are not eligible for concessional financing by the IMF at the time of program approval. This excludes low-income countries (LICs), for which market access matters less, as they primarily rely on foreign aid as a source of external finance.

We measure geopolitical interests with a variable on temporary membership in the United Nations Security Council (UNSC), following the approach used in Dreher et al. (2009). Temporary UNSC membership makes a country exposed to US geopolitical interest, as the US can incentivize the country to vote in line with the US on the Security Council. Given the strong influence of the US on IMF policy making, it can for example reward the country by ensuring softer IMF lending conditions, which reduces the political cost of IMF lending for the country in question.

Using ordinary least squares (OLS), we analyze the effect of a new IMF program while being temporary member of the UNSC on four different financial market variables: sovereign bond yields, yields of short-term government bills, domestic stock prices, and exchange rate movements towards the US dollar.

We find that geopolitical interest in a country that receives an IMF program tend to increase risk aversion of financial markets in the short-term, or investor selling of the country’s assets, compared to countries that are not in the focus of geopolitics. This reduces a potential catalytic effect of IMF programs, which might render IMF programs less effective. Specifically, our finding is that for temporary UNSC members, a new IMF program is consistently associated with a negative reaction by financial market participants for four financial market variables. For government bonds and short-term bills, a new IMF program during temporary UNSC membership is associated with a sizeable increase in yields, indicating that investors sell bonds and bills. At the same time, domestic stock price development stalls as investors sell them, and the domestic exchange rate to the US dollar depreciates as investors sell domestic currency. We do not find such a consistent negative reaction by financial market participants for program approvals for countries that are not temporary UNSC members. While many of these results cannot be estimated very precisely, they overall suggest that geopolitics reduce the catalytic effect of IMF programs, making them less effective.

We add to the existing literature by analyzing the role of geopolitics on the IMF catalytic effect using temporary membership in the UNSC. This builds on a rich body of literature on the link between geopolitics and IMF lending, which we summarize in the next section. Specifically, we base our research on Dreher et al. (2009), who study the effect of UNSC temporary membership on conditionality in IMF programs. The UNSC variable has the advantage of being quasi-random, and hence addresses the potential endogeneity issues of other variables used in earlier research. Our empirical approach builds on Chapman et al. (2017), who analyze the effect of IMF lending announcements on government bonds in the context of geopolitical interests. However, instead of using annual data, we are the first (to our knowledge) to use and analyze a large monthly panel data set in this context.³ This allows a much more precise, albeit short-term, study of the reaction of financial markets. We further extend the literature by enlarging the variables used to measure financial market reactions to include not only bonds, but also short-term bills, stock prices, and exchange rates, while expanding the data by covering the period January 1993 until December 2019 and about 100 IMF member countries.

The rest of this paper is structured as follows. Section “Literature review and theoretical considerations” gives an overview of the extensive literature on politics affecting economic outcomes of official financial flows, on the more specific the link between geopolitics and IMF lending, and on the IMF’s catalytic effect. The section further offers some theoretical considerations on the impact of geopolitical interests on the financial market reaction to IMF programs. Section “A closer look at Pakistan, Brazil and Peru” allows a closer look at the data for three specific cases, Pakistan, Brazil and Peru. Section “Data and descriptive statistics” presents the data used in the empirical analysis together with some descriptive evidence. Section “Empirical strategy” lays out the empirical strategy, while Section “Results” presents the results. The last section concludes.

Literature review and theoretical considerations

Political interests and the economic outcomes of official financial flows

The analysis of how geopolitics affect the financial market reactions to IMF programs, which is the focus of this paper, can be put in a broader context. Geopolitical interests are a specific form of political interests in general, and IMF lending can be considered a special form of ‘official’ financial flows, i.e. flows between governments. How financial markets react to an economic reform program is likely affecting its program’s outcome. Hence, our research question addresses a specific situation in the general debate about whether international official financing is less effective in achieving its economic objectives when it is provided for political reasons.

As Dreher et al. (2013) point out, a key tenet of the realist school of international relations theory is that governments give money to other governments for political goals, while at the same time accepting inferior economic outcomes. This topic has been extensively researched.⁴ One strand of the literature focuses on the failure of development assistance to make a meaningful impact on economic growth, which Alesina and Dollar (2000), in their seminal study on the subject, similarly attribute to political interests influencing aid allocation. It can be argued that if aid flows are influenced by political interest, they are less likely to improve economic conditions in the recipient country but only increase unproductive public consumption, linked to poor institutional development and corruption (Alesina & Dollar, 2000, p. 33).

As a further addition to the literature, Dreher et al. (2013) find that the performance of World Bank projects, i.e. official World Bank aid flows, declines when allocated to countries in challenging economic situations for geopolitical reasons. Dreher et al. (2018a) find that politically motivated aid is insignificant or even harmful to growth. Malan (2018) investigates how having a country representative at the IMF Executive Board may allow the country to influence the amounts and repayment terms of IMF lending it receives. Dutta and Williamson (2019) look into the nuances of political incentives of donors in aid allocations and point to the challenge of endogeneity with regards to aid flows. Dreher et al. (2021) investigate the effect of Chinese aid flows on economic development at the subnational level and find no negative relation. Berlin et al. (2023) look at domestic politics and find further evidence that World Bank aid can be exchanged by recipient governments for favors that are of geopolitical value to the major powers.

How political interests could undermine the IMF’s catalytic effect

The role of political considerations in IMF lending is closely related to the more general debate outlined above. If political interests become too important in the IMF’s efforts in stabilizing countries via its lending activities, this could weaken the credibility of the IMF’s ability to fulfil its task of ensuring global macroeconomic stability (IEO, 2002).

From a theoretical perspective, a key aspect of an IMF program for a country in economic crisis is its signal to financial market participants that the country has adopted adequate economic policies to overcome the crisis. This signal relies on the assumption that the IMF’s reform advice is based solely on the IMF’s interest to stabilize the economic situation of the country in question. According to the IMF, this so-called catalytic effect is a major goal of its lending, as ‘IMF programs can help unlock other financing, acting as a catalyst for other lenders. This is because the program can serve as a signal that the country has adopted sound policies, reinforcing policy credibility and increasing investors’ confidence’ (International Monetary Fund [IMF], 2020b). Hence, we assume that all other things being equal, IMF programs are ‘good news’ for financial market participants.

However, if the IMF’s decision to lend to a country is perceived as not solely based on sound economics, but also influenced by political considerations, we assume that this reduces the catalytic effect of the program, as it might be less credible from the perspective of achieving economic stabilization in the country in question. For example, political interests in achieving agreement on an IMF program may lead to less focused or stringent reforms as part of the program, or less monitoring by the IMF of whether the stipulated reforms are actually implemented.

There is a broad literature on the question whether the assumed catalytic effect of IMF lending on financial markets actually can be observed.⁵ Many studies find mixed evidence, such as Bird and Rowlands (2002) and (2008), Edwards (2006), and Eichengreen et al. (2008). At the same time, Mody and Saraiva (2006) find evidence for the existence of an IMF catalytic effect, but only in cases where financial market participants perceived the planned reforms under the arrangement as credible. Based on newer data, Krahnke (2020) finds further evidence of the existence of an IMF catalytic effect, but only up to a certain IMF program size. Gehring and Lang (2020) find that financial market participants perceive IMF programs as a positive signal, despite a contractionary effect on the economy.

Literature on the influence of geopolitics on how the IMF lends

While there is a large amount of literature on how political interests in a broader sense influence IMF lending decisions,⁶ Thacker (1999) provides the first systematic evidence of the more narrow focus on the role of geopolitical interests in this context. Concentrating on US geopolitical interests, he finds that since the end of the Cold War, political proximity to the US, as measured by United Nations voting patterns, has increased the likelihood of receiving IMF programs. Stone (2004) adds evidence that US political influence is linked to poorer results of IMF lending in Africa during the 1990s.

US political interests may not always be of geopolitical nature but can also relate to financial interests. In this vein, Oatley and Yackee (2004) find that US financial interests in a country asking for an IMF program lead to a larger loan size and fewer conditions. Broz and Hawes (2006) find further evidence for a positive link between the size of an IMF program and foreign bank exposure in the program country.

Vreeland (2005) reconciles contradictory findings on the role of domestic and international policy interests. He argues that governments use IMF programs to push through with unpopular reforms, but only in the absence of US geopolitical interests. Similarly, Sturm et al. (2005) address the contradicting evidence on the influence of political factors in IMF lending in earlier research. They find that the majority of political variables suggested in earlier research are not significantly related to two widely used dependent variables, which are the signing of an IMF arrangement and the use of IMF credit. In particular, while political factors influence the signing of an IMF arrangement, IMF lending activities are primarily related to economic factors.

Stone (2008) finds that US geopolitical interests are linked to conditions in IMF programs in fewer areas, such as monetary policy, fiscal aspects, or financial stability—hence, a more limited scope of conditions. Dreher et al. (2009) investigate the link between IMF programs and a newly proposed measure of US geopolitical influence - temporary UNSC membership. They find a robust positive relationship between UNSC membership and IMF program participation, as well as evidence that UNSC membership reduces conditionality in IMF programs. Dreher et al. (2018b) further find that voting in line with or against the US in the UNSC is related to the likelihood of signing of an IMF arrangement in a given year and to IMF loan size.

Biglaiser and DeRouen (2010) and Woo (2013) look more specifically at how US business interests are associated with IMF programs. Biglaiser and DeRouen (2010) find that IMF programs can benefit US firms abroad, such that IMF borrowing countries attract more US foreign direct investment (FDI) than countries that do not borrow from the IMF. Similarly, Woo (2013) analyzes how the design of IMF programs affects FDI flows to the borrowing country. He finds that stricter conditionality in an IMF program is associated with higher FDI inflows, as more conditions signal a greater commitment to economic reforms by the borrowing government.

Moser and Sturm (2011) update and extend the work by Sturm et al. (2005) and find a number of non-economic variables to be robustly related to IMF programs. These are the existence of a history of IMF programs in the country prior to the current program, domestic elections around the program start, how autocratic a country is, the country’s level of international reserves, real economic growth rates, and currency crises. However, they do not find a robust positive link between IMF lending and geopolitical interest variables such as UNSC membership and US trade relations.

Dreher et al. (2015) analyze the effect of temporary membership in the UNSC on IMF conditionality. They use dependent variables on the number of conditions and the scope of conditions, i.e. in which area of the economy the conditions apply. They find that UNSC members face about 30% fewer conditions per program. Regarding the scope of conditions, they find that UNSC temporary membership is linked to fewer conditions in crucial areas of the economy, such as debt repayment, balance of payments, credit to the government, and domestic pricing.

Finally, a number of authors move beyond the interests of the US and investigate the influence of major powers in IMF lending more generally. Copelovitch (2010) finds empirical support for a common agency of powerful countries in IMF lending. He shows that if aggregate interests are weak, bureaucratic and technocratic interests of IMF staff influence IMF loans. Presbitero and Zazzaro (2012) find that political similarity with the G7 has a positive effect on the probability of signing an IMF program in a given year. Breen (2014) further argues that major powers cooperate to reduce conditionality in an IMF program if their banks and exporters are exposed to the requesting country, and would risk loss in the event of IMF payout suspension due to not meeting the imposed conditions.

Measuring geopolitical interest

A key question in the empirical research on geopolitics and IMF lending is how to measure geopolitical interest empirically. In this study, we will follow the approach by Dreher et al. (2009) to focus on temporary membership in UNSC.⁷ Temporary member states of the UNSC may trade their UNSC votes to the US for better accessibility of IMF loans, which they value higher than their UNSC voting power. UNSC temporary membership thus appears to be a good measure for US geopolitical interests. The variable has the benefit of being close to exogenous, which we explain in more detail in Section “Endogeneity issues”.

Dreher et al. (2009) find that UNSC membership makes an IMF program more likely. Dreher et al. (2015) add evidence that links UNSC temporary membership to fewer conditions in crucial areas of the economy, such as debt repayment, balance of payments, credit to the government, and domestic pricing.

Do geopolitical interests affect the financial market reaction to IMF programs?

The aim of this study is to investigate the link between geopolitical interests in IMF lending and the catalytic effect of IMF programs. As laid out above, we assume that geopolitical interests in IMF lending reduce the catalytic effect, as their presence renders financial markets less convinced that a reform program achieves economic stabilization in the country in question. This is in line with the general assumption that political considerations in the allocation of official financial flows leads to worse economic outcomes.

Chapman et al. (2017) argued that geopolitical motivations affect the catalytic effect of an IMF program, focusing on the effect on government bonds. Building on their research design, our key contribution is that we address possible endogeneity issues by using a quasi-exogenous variable to measure geopolitical influence. Following Dreher et al. (2015), we use temporary membership in the UNSC as a measure of US geopolitical interest. Our second major contribution is that we are (to our knowledge) the first to analyze monthly instead of yearly data in the context of IMF programs and geopolitics. The use of higher frequency data allows a much more precise, albeit shorter-run, measurement of the impact of a new IMF program on financial markets. Furthermore, compared to prior research, we use a considerably longer data period spanning 27 years, and expand country coverage by including about 100 IMF member countries. We use more variables to measure the reaction by financial markets, by including bills, stocks, and exchange rates. We also expand the data coverage on bonds by including JPM emerging market bond indices. Finally, we expand the set of controls. In addition to interacted year-country dummies, which should capture most country-level factors changing on a yearly level, we add monthly inflation. We also include dummies for each time observation, i.e. month-year, to capture overall time trends in a flexible way.

A closer look at Pakistan, Brazil and Peru

To get a better understanding of the effect of temporary UNSC membership on IMF lending, we take a closer look at three country cases, Pakistan, Brazil and Peru. These three are examples of countries receiving IMF programs while being temporary UNSC members.

In the case of Pakistan, the data suggest a clear negative reaction by financial markets to the IMF program approved in 2013. This supports our hypothesis that geopolitics reduce the IMF catalytic effect. However, it can also be argued that in the last decades, Pakistan was almost permanently of geopolitical importance to the US—hence, temporary UNSC membership might not necessarily make a difference. Therefore, we add a second case study that shows the expected negative market reaction - Brazil. Financial markets reacted negatively to Brazil’s IMF program in 1998, a time when the country was also temporary UNSC member. In both cases, the financial market reaction peters out after a few months. This is in line with usually observed financial market reactions to news, as other ‘new’ events quickly take over in importance. Finally, we investigate a third case study, which does not appear to fit our hypothesis—Peru’s IMF program of 2007.

Pakistan

There is ample anecdotal evidence that US geopolitical considerations strongly influenced Pakistan’s many IMF programs over the last decades—not only in the case of the IMF program for 2013, which we focus on in this study. IMF lending to Pakistan is connected to the on and off political relations between the US and Pakistan (Ahmad & Mohammed, 2012). At the core of US interests in Pakistan is the country’s strategic location at the Persian Gulf, which is key in the US war on terror. According to a 2002 report of the IMF IEO, there is ample anecdotal evidence, based on interviews with Pakistani authorities and IMF staff, that many programs primarily served political considerations. There appeared to be a sense that due to geopolitical interests, the IMF would support Pakistan irrespective of program implementation or success (IEO, 2002, p. 131).

The IMF lending to Pakistan that is interesting for this study is the program of 2013. Pakistan was elected to the UNSC as temporary member in October 2011. It served on the council from January 2012 until December 2013. In September 2013, Pakistan received a three-year program under the IMF’s Extended Fund Facility (EFF). The aim of the IMF program was to support Pakistan’s economy in the context of Taliban violence and deep-rooted corruption. The program should also support the reform efforts of the newly elected Prime Minister at the time, Nawaz Sharif, by providing some economic stability (Walsh & Masood, 2013).

Financial markets reacted negatively to the new program, with rising bond yields, lower stock price development, and exchange rate depreciation (see Figure 1). In fact, the yields of bonds and bills started to increase already a few weeks before the program approval. The same is true for the exchange rate, which slowly depreciated over the summer months. Indeed, the IMF published information about the ongoing program negotiations in July (Walsh & Masood, 2013). In the months after the program approval, bills and bond yields stayed at an elevated level, while the exchange rate and stock prices eventually recovered.

Figure 1. Pakistan: Negative short-term reaction of financial markets to IMF program approval while being temporary member of the United Nations Security Council (UNSC).

Note: Chart shows the short-term reaction of financial market indicators to a new IMF program (vertical blue line) for countries that are temporary UNSC members. The time range shown is IMF program approval months plus/minus 12 months. The chart is constructed so that a falling line shows a negative reaction by financial markets. The exchange rate is against the US dollar; rhs = right-hand scale.

Negotiations about facilitating access to IMF programs, such as in exchange for UNSC votes, usually take place behind closed doors. Hence, it is not surprising that there is no direct evidence of a link between the IMF program approval and Pakistan serving on the UNSC. However, the timing of the two events is right to allow a connection. There is also ample evidence for strong US geopolitical interest in Pakistan during the period in question, mainly in the context of US war efforts in Afghanistan. In 2009, the US officially adopted a policy approach to Pakistan that would support its strategic ties by ample, economic aid (Kronstadt, 2015). In 2011, relations temporarily soured in the context of the US killing of Osama bin Laden in Pakistan. Relations improved again in 2012, when both sides agreed to renew their cooperation in the interest of stability in the region, which was vital for US interests in Afghanistan. These renewed relations were supported by resumption of considerable US aid flows to Pakistan, which had been set on hold in 2011. The US welcomed the election of Nawaz Sharif in 2013, and a few months later, the two countries formally announced their interest in deepening their partnership (Rafique, 2015). Hence, while Pakistan had been in the focus of US geopolitical interests already before 2013, as described above, the 2013 IMF program nevertheless falls into a special moment of US-Pakistani relations from a geopolitical perspective.

In conclusion, it seems very likely that US geopolitical interests influenced the IMF program of 2013 for Pakistan. In line with our hypothesis, financial market participants reacted with increased risk aversion to the program and sold Pakistani financial assets. This likely reduced the catalytic effect of the IMF program on capital flows.

Brazil

Brazil received the approval for an IMF program in December 1998 to avert a currency crisis related to the financial crises in Asia and Russia at the time. At the same time, Brazil was elected to the UNSC as temporary member in October 1997 and served on the council from January 1998 to December 1999.

Looking at the financial market data at the time of the program approval, markets showed a clear negative reaction, with bond and bill yields rising, the exchange rate depreciating, and stock prices falling (see Figure 2). The timing of the reaction seems to coincide with the Board approval date in December. There were leaks to the press about the new program shortly ahead, in mid-November,⁸ but the monthly data likely blurs this. The negative financial market reaction continued a few months after program approval and then stabilized.

Figure 2. Brazil: Negative short-term reaction of financial markets to IMF program approval while being temporary member of the United Nations Security Council (UNSC).

Note: Chart shows the short-term reaction of financial market indicators to a new IMF program (vertical blue line) for countries that are temporary UNSC members. The time range shown is IMF program approval months plus/minus 12 months. The chart is constructed so that a falling line shows a negative reaction by financial markets. The exchange rate is against the US dollar; rhs = right-hand scale.

Brazil’s IMF lending history shows that the country had only one other program in the 1990s, from 1992 to 1993. Thus, the 1998 program was not merely a continuation of previous lending. In addition, the agreed amount of 41.5 billion US dollars for the three-year program was several times the size of earlier programs, which usually had lasted only one year.

Vreeland (1999) analyses the 1998 IMF program for Brazil in view of the influence of political interests. He argues that domestic political interests were an important incentive for the IMF program, more than the immediate economic need. The program was meant to help the Brazilian government push through unpopular reforms, with ‘the whole world watching’ (p. 3).

Brazil’s 1998 program was also important to US economic interests. The US wanted to avoid a potential regional destabilization from economic crisis in Brazil, which could also negatively affect US companies exposed to the region (Watkins, 2009). In this vein, the US added a bilateral loan of 5 billion US dollars to the lending package. Thus, the US showed strong support for a program whose success chances where highly debated (Blustein, 1998).

More generally, the US and Brazil have a history of robust political and economic relations, but not always without difficulties due to differing national interests. Difficulties arose particularly around Brazil’s economic policies in the context of the Mercosur common market, which the US perceived as protectionist, and Brazil siding with other emerging markets against the US in WTO negotiations. Security relations and military ties between the countries were more limited at the time in question (Meyer, 2020).

In summary, while there is no straightforward evidence for a quid-pro-quo in view of US geopolitical interests in this IMF program, there is also no indication to the contrary. The financial market reaction clearly supports our hypothesis that financial market participants reacted with increased risk aversion to the program and sold Brazilian financial assets.

Peru

Peru was elected to the UNSC as temporary member in October 2005. It served on the council from January 2006 until December 2007. In January 2007, Peru received a two-year Stand-by Arrangement (SBA) for 270 million US dollars. This program was purely precautionary, and Peru never actually drew the money. Importantly, this was not a ‘one-off’ program to fight a specific economic crisis. Instead, it was the last of seven consecutive IMF programs, with program length varying between one to three years, which Peru received between 1993 and 2009. After 1999, Peru stopped drawing on the programs, and they became purely precautionary.

The SBA of 2007 was relatively small compared to earlier programs. It can thus be argued that this program was considered an insurance-like program for Peru. Investors likely perceived the rolling over of Peru’s IMF programs as established routine. Given that Peru had received so many IMF programs before 2007, it seems unlikely that financial markets would think that the 2007 program was only due to Pakistan’s UNSC membership at the time. Hence, it is questionable whether there was any meaningful financial market reaction to the IMF program approval in January 2007.

Looking at the financial market indicators around the IMF program approval (see Figure 3), the exchange rate continued to appreciate slightly. Stock prices in US dollars continued their upward trend, but at an even faster pace after the program approval. Bond yields, however, do not show a clear trend. Overall, the financial market reaction seems neutral or slightly positive, but rather muted.

Figure 3. Peru: Positive short-term reaction of financial markets to IMF program approval while being temporary member of the United Nations Security Council (UNSC).

Notes: Chart shows the short-term reaction of financial market indicators to a new IMF program (vertical blue line) for countries that are temporary UNSC members. The time range shown is IMF program approval months plus/minus 12 months. The chart is constructed so that a falling line shows a negative reaction by financial markets. The exchange rate is against the US dollar; rhs = right-hand scale.

From a geopolitical perspective, Peru and the US enjoyed close relations during the time in question, with a special cooperation in the area of counternarcotic and security matters (Jasper & Ribando Seelke, 2008). The US was also the key trading partner of Peru at the time, and the two countries signed a Trade Promotion Agreement in April 2006 (Taft-Morales, 2009).

Overall, despite close relations to the US, in view of Peru’s long history of IMF programs it is hard to argue that the SBA of 2007 was a particular reward for voting in line with the US in the Security Council. Rather, the 2007 program seems to be a logical continuation of a historical pattern of consecutive IMF programs for Peru, and on which the Peru’s temporary UNSC membership did not have a major influence.

Data and descriptive statistics

Dataset

To test whether there is general evidence in favor of our hypothesis, we analyze a panel data set of country-level monthly data from January 1993 to December 2019, for about 100 IMF member countries, depending on the sample. The time period is defined by the IMF MONA database, which is available from 1993 onwards.⁹ While we start by including all IMF member countries, we use a dummy to ensure that only countries enter the sample that are not eligible for concessional financing by the IMF at the time of program approval. The reason is that countries eligible for concessional financing by the IMF’s Poverty and Reduction Growth Trust (PRGT) are poorer countries, or LICs, which largely rely on official aid flows as a main source of external financing and usually have no real access to global capital markets. Therefore, the financial market indicators, such as the exchange rate, are dominated by other drivers than in the case of mid- and high-income emerging markets.

In contrast to the majority of literature (with the notable exception of Gehring and Lang, 2020), we use a monthly periodicity and thereby allow the study of short-term financial market reactions. This is not possible when using annual data as done in almost all other studies on IMF lending. We abstain from using a daily dataset, as we do not consider changes at a daily horizon to be very relevant for policy and lending decisions. Furthermore, daily data is likely to become rather noisy, reducing the likelihood for us to pick up the actual and more persistent signals of financial markets.

Variables¹⁰

Dependent variables

As dependent variables, we study four financial market variables that are widely used for emerging market economies on a monthly basis: sovereign bond yields, short-term bill yields, domestic stock prices in local currency, and the exchange rate to the US dollar. For all of these variables, there is an established reaction if investor sentiment turns positive or negative towards the country in question.¹¹

Sovereign bonds of emerging markets are generally considered relatively risky compared to bonds of advanced economies, which is reflected in overall higher yields. If investors have a positive sentiment towards a country’s financial markets, such as due to the suggested catalytic effect of a new IMF program, they will buy bonds, leading to a lower yield. In a negative reaction, investors will sell emerging market bonds, leading to lower bond prices and higher bond yields. The bond variable in this study reflects the yields on government bonds of the country in question. The variable is based on the JPM EMBI bond index, complemented by the IMF IFS database in case of gaps. The maturity of the included bonds is at least 2.5 years. To improve distributional characteristics, the variable is transformed by 100*x/(x + 100) and subsequently winsorized.¹²

For government short-term bills (usually three-month treasury bills), the pattern of financial market reaction is similar to bonds but more pronounced because of shorter maturities. If investors sell bills (in a negative financial market reaction), the price of bills will fall, and yields will increase. The variable on bills in this study reflects the yield (in percent) of short-term treasury bills of the country in question, based on the IMF IFS database. To improve distributional characteristics, the variable is transformed by 100*x/(x + 100) and subsequently winsorized.

Emerging market stocks are considered a high-risk investment. If investor sentiment turns negative, investors will sell emerging market equities, which leads to lower stock price developments. The stock-market variable we use is the monthly stock price index in local currency of the country in question, based on MSCI country indices. To improve distributional characteristics, month-over-month growth rates are taken, and the variable is subsequently winsorized.

Finally, the domestic exchange rate to the US dollar is likely to be one of the most quickly reacting financial assets in case of changing investor sentiment. However, this is only the case if the exchange rate is floating, which means that the central bank of the country does not intervene heavily in the exchange rate market. Hence, we only focus on cases where the exchange rate regime in the given month is considered ‘floating’ by the IMF in its Annual Report on Exchange Rate Arrangements and Exchange Restrictions (AREAER) (IMF, 2020a). Exchange rate data is based on the IMF IFS database. To improve distributional characteristics, month-over-month growth rates are taken and subsequently winsorized.

Variable of interest

The variable of interest in this study is the interaction term between a dummy on a new IMF program and a dummy on UNSC temporary membership. It measures the impact of UNSC membership on the IMF program effect and thereby proxies whether the catalytic effect put forward by the IMF is affected by exogenously driven geopolitical interest.

The dummy on a new IMF program is constructed to take the value of 1 in the month when a new IMF program is approved by the IMF Executive Board, and 0 in all other cases. The source is the IMF MONA database. IMF program negotiations normally take place behind closed doors, so we assume that IMF programs are a surprise for financial market participants until they first hear about it. This is in most cases the day of the press release about the new program, which is normally the same date as the IMF board approval. In case of earlier leaks to the press, the date when financial markets learn about the new IMF program can also be several weeks or months earlier, such as in the case of Pakistan (see Section “Pakistan”). As the IMF Executive Board approval dates are available for all cases, we use this date for consistency.

The dummy variable on UNSC takes the value of 1 during the election and serving time of a country as non-permanent member of the UNSC, and the value of 0 otherwise. Mostly, the election takes place in fall of a given year, and the time of serving on the council starts in January of the following year. A non-permanent member usually serves for two years, but shorter periods are possible. The dummy used in this study takes the value of one in the election month and remains one until the end of the serving time, so about 26-27 months in total. Hence, we cover the entire period in which a country is a non-permanent member of the UN Security Council, or is known to be about to become one.¹³

Controls

An ideal set of controls would be based on the list of economic and political variables found to be the most robust determinants of IMF programs in past literature, such as by Moser and Sturm (2011, p. 325). These include, for example, measures of economic growth and the debt situation, information about past and upcoming domestic elections, as well as data on the economic, financial and political globalization of the country in question. However, very few of such economic or political variables are widely available for emerging markets on the monthly basis that we use for our analysis (most past research uses yearly data, as described above). As described in the empirical strategy in Section “Empirical strategy”, we address this issue by including monthly dummies to cover global developments that affect all countries, and year-country interaction dummies to cover country-level developments changing on a yearly level.¹⁴ Hence, we concentrate on the within-country, within-year variation.

In addition, we add country-level inflation data, which is the only macroeconomic indicator available for a large group of countries on a monthly basis. Changes in inflation are associated with a country’s financial market variables, such as higher bond and bill yields, lower stock price development, and exchange rate depreciation. We base this variable on the monthly country-level Consumer Price Index (CPI) as a percentage change from the previous year, based on national sources. To improve distributional characteristics, the variable is transformed by 100*x/(x + 100) and subsequently winsorized.

Some descriptive evidence

Before advancing to the econometric analysis of the data, we offer some descriptive evidence on the key variables we analyze, the financial market variables and the two dummies on a new IMF program and on temporary UNSC membership.

Table 1 lists the 14 cases in the dataset in which a country receives an IMF program while being elected to serve or serving as a temporary member on the UNSC. Overall, there are 233 observations of new IMF program approvals in the dataset, and 109 cases in the dataset where a country is elected to the UNSC as non-permanent member in a given month.

Table 1. Cases of new IMF program approvals for countries that are at the same time temporary members of the United National Security Council (UNSC).

Country	IMF program approval	UNSC election	UNSC serving time
Korea	December 1997	November 1995	January 1996 – December 1997
Brazil	December 1998	October 1997	January 1998 – December 1999
Argentina	March 2000	October 1998	January 1999 – December 2000
Bulgaria	February 2002	October 2001	January 2002 – December 2003
Romania	July 2004	November 2003	January 2004 – December 2005
Peru	January 2007	October 2005	January 2006 – December 2007
Costa Rica	April 2009	November 2007	January 2008 – December 2009
Mexico	April 2009	November 2008	January 2009 – December 2010
Mexico	March 2010	November 2008	January 2009 – December 2010
Colombia	May 2011	October 2010	January 2011 – December 2012
Portugal	May 2011	October 2010	January 2011 – December 2012
Morocco	August 2012	November 2011	January 2012 – December 2013
Pakistan	September 2013	November 2011	January 2012 – December 2013
Egypt	November 2016	October 2015	January 2016 – December 2017

Figure 4 offers a visualization of the average values of the financial market variables for all cases (orange bars) and for IMF program cases, differentiated by UNSC temporary membership at the same time (blue bars without, grey bars with UNSC membership). For all four financial market variables, the latter two means are not statistically different from each other.

Figure 4. Average values for financial market variables for 1993-2019.

Notes: Orange bars show the averages from 1993-2019 for yields of bonds and bills, growth in domestic stock prices (in local currency), and growth in the exchange rate to US dollars (if considered floating). Blue bars show the averages in case of a new IMF program for countries that are not temporary members of the United Nations Security Council (UNSC). Grey bars show the average in case of a new IMF program and contemporaneous temporary UNSC membership. Numbers of observations are a shown above the bars.

Overall, the average bond yield across the sample is 6.3% (see left upper chart in Figure 4). In presence of a new IMF program, and without UNSC membership (which is the majority of cases), the average bond yield is higher at 8.9%, which suggests a negative financial market reaction instead of a catalytic effect of an IMF program. In the presence of temporary UNSC membership, the negative reaction persists but seems a bit lower, with an average bond yield of 8.4%.

The average yield of short-term bills (right upper chart in Figure 4) across the dataset is 7.5%. For IMF program cases, without UNSC membership, the bill yield is considerably higher, at 11.8%, again suggesting a negative financial market reaction. For new IMF programs while being temporary UNSC members, the bill yield increase is again close to the entire dataset (7.9%).

The average stock price growth (in local currency, lower left chart in Figure 4) in the dataset is 0.7. The growth rate is considerably higher for IMF program cases (1.9), and even more so if the country is UNSC temporary member at the same time (2.4). This suggests a very strong catalytic effect of IMF programs on stock prices. However, it is possible that instead of positive investor sentiment, the exchange rate effects drive these increases, as the exchange rate chart shows (see chart on the lower right). The average exchange rate growth for all cases in the dataset (local currency against the US dollar, hence depreciation) was 0.5. For the cases of IMF programs, it was very similar. However, for IMF programs while being UNSC temporary members, countries saw a much higher depreciation, at 1.1.

Empirical strategy

Estimation strategy

Our hypothesis is that a new IMF program while being temporary UNSC member is overall associated with a negative reaction of financial market participants. For simplicity and transparency, we use an OLS linear regression model with robust standard errors. We thus test (1) $$\begin{array}{c}{Y}_{\mathit{\text{it}}}={\beta }_1{\mathit{\text{IMFNEW}}}_{\mathit{\text{it}}}{\times \mathit{\text{UNSC}}}_{\mathit{\text{it}} }+{\beta }_2{\mathit{\text{IMFNEW}}}_{\mathit{\text{it}}}{+{\beta }_3\mathit{\text{UNSC}}}_{\mathit{\text{it}}}\\ +{{\beta }_4{\mathit{\text{CPI}}}_{\mathit{\text{it}}}+{\beta }_5\delta }_{\mathit{\text{year}}}\times {\delta }_i+{{\beta }_6\delta }_t+{{\beta }_7\delta }_i+u_{\mathit{\text{it}}},\end{array}$$(1) where $Y_{\mathit{\text{it}}}$ is the dependent variable. This is, alternatively, the sovereign bond yield, short-term bill yield, the growth rate of domestic stock price in US dollars, or the growth rate of the domestic exchange rate against the US dollar (provided it is considered ‘floating’ by the IMF at that point in time) for country i in month t.

The variable of interest is ${\mathit{\text{IMFNEW}}}_{\mathit{\text{it}}}{\times \mathit{\text{UNSC}}}_{\mathit{\text{it}}}$, which is the interaction term between a dummy for a new IMF program approval and a dummy on temporary UNSC membership for country i in month t. Based on our hypothesis, we expect a positive sign for bonds and bills (indicating an increase in yields), a negative sign for stocks (indicating falling prices), and a positive sign for the exchange rate (indicating depreciation). The coefficients of ${\mathit{\text{IMFNEW}}}_{\mathit{\text{it}}}$ and ${\mathit{\text{UNSC}}}_{\mathit{\text{it}}}$ cover their effects on the dependent variable when these do not occur simultaneously. Hence, the interaction term ${\mathit{\text{IMFNEW}}}_{\mathit{\text{it}}}{\times \mathit{\text{UNSC}}}_{\mathit{\text{it}}}$ does only show the additional effect of having both situations occur at the same time. It might be that some of the new IMF programs while being temporary UNSC member would have also been granted in case the country would not have been a temporary member. In that case, the effect on the respective financial market variable should not be any different from that captured by the ${\mathit{\text{IMFNEW}}}_{\mathit{\text{it}}}$ variable itself. To the extent that this occurs, this therefore implies a reduced (absolute) size of the coefficient estimate in front of the interaction term. Put differently, if we would be able to exclude such situations from our analysis, then the impact of the truly politically driven support triggered by the temporary UNSC membership would likely be larger. By design, we are looking for an average effect, which might imply an underestimate of the consequences of a truly politically motivated IMF program.

${\mathit{\text{CPI}}}_{\mathit{\text{it}}}$ is inflation for country i in month t. An interaction term ${\delta }_{\mathit{\text{year}}}{\times \delta }_i$ of year and country dummies captures country-specific factors subject to yearly changes, such as domestic economic and political conditions. ${\delta }_t$ is a monthly dummy that captures global developments on a monthly level that affect all countries, such as the global financial crisis. The $\beta$ vectors capture the effects of these variables. Finally, $u_{\mathit{\text{it}}}$ is the error term. We use standard errors clustered by country, as we expect the residuals within countries to be correlated.¹⁵

Endogeneity issues

The ideal experiment would be to assign to each country an IMF program randomly, and to assign to the country some geopolitical interest, measured by UNSC temporary membership, in an equally random manner. The variable on UNSC temporary membership, which we use to measure geopolitical interest, is quasi-random, as explained in detail in Dreher et al. (2015, p. 125). The ten temporary seats of the UNSC are not a random draw but they are allocated by region, with different regions following different selection processes, mostly based on turn taking. While Africa has the strongest turn-taking, regional hegemons will dominate in other areas of the world. No strong pattern seems to exist for Eastern Europe. The nominations for temporary membership are agreed by the regions, and then ratified by the United Nations General Assembly, with competitive elections by two-third majority of the Assembly taking place in about 20% of the cases. The term limit of two years reinforces the exogeneity of the selection process (Dreher et al., 2014).

A new IMF program is not random but related to a multitude of economic and political factors, most obviously an economic crisis. We try to capture most of them by applying not only a control on inflation, but also an interaction term of year and country dummies, as described above. At the same time, a new IMF program is unlikely to be related to UNSC temporary membership given the quasi-random nature of that variable.

Finally, it is important to point out that financial market conditions very likely impact the likelihood of an IMF program. For example, worsening risk appetite by international financial market participants may trigger capital outflows of a country, which can lead to severe economic crisis and need for IMF support in the case of underlying vulnerabilities. We account for this by using monthly dummies and country-year dummies as controls, which should largely capture domestic and international financial market conditions.

This said, we do need to realize that our coefficient of interest is the one in front of the interaction between temporary UNSC membership and receiving a new IMF program (the ${\beta }_1$-coefficient in Equation ((1))(1) $$\begin{array}{c}{Y}_{\mathit{\text{it}}}={\beta }_1{\mathit{\text{IMFNEW}}}_{\mathit{\text{it}}}{\times \mathit{\text{UNSC}}}_{\mathit{\text{it}} }+{\beta }_2{\mathit{\text{IMFNEW}}}_{\mathit{\text{it}}}{+{\beta }_3\mathit{\text{UNSC}}}_{\mathit{\text{it}}}\\ +{{\beta }_4{\mathit{\text{CPI}}}_{\mathit{\text{it}}}+{\beta }_5\delta }_{\mathit{\text{year}}}\times {\delta }_i+{{\beta }_6\delta }_t+{{\beta }_7\delta }_i+u_{\mathit{\text{it}}},\end{array}$$(1) ). According to Dreher et al. (2018a) working in a relatively comparable setting and referring to Bun and Harrison (2019) and Nizalova and Murtazashvilli (2016), this coefficient can be estimated consistently under basically two assumptions that do not require receiving an IMF program to be exogenous. The first is the exogeneity of temporary UNSC membership, conditional on the variables of the model. The second is that the endogeneity of an IMF program due to an omitted variable bias must be independent of UNSC status, i.e. we need to assume that a bias due to the (potential) endogeneity of a new IMF program does not depend on whether a country is a temporary member of the UNSC.

Results

The results section is divided in two parts. We first shed light on the first-order effects of the two key dummies on a new IMF program and on UNSC temporary membership on the financial market variables separately. We then present the main results based on the interaction effect of the two variables. Our main result is that a new IMF program is consistently associated with a negative reaction of financial market variables if the country also is a temporary UNSC member. This is not the case for new IMF program approvals when a country is not a temporary UNSC member.

First-order effects

In a first step, we provide the individual results for two key dummies on a new IMF program and on UNSC temporary membership on the financial market variables. The short-term reaction of financial markets to a new IMF program is limited and statistically insignificant at conventional confidence levels (see Table 2, first row), but they suggest a small catalytic effect of IMF lending in the short term. In reaction to a new IMF program, bonds hardly move, while bill yields slightly decrease, stock price development is positive, and the exchange rate tends to appreciate.

Table 2. Financial market reaction to new IMF program and United Nations Security Council (UNSC) temporary membership.

	(1) Bond yield	(2) Bill yield	(3) Stock prices (growth rate, local currency)	(4) Exchange rate (growth rate, to US dollar)
New IMF program	0.0982	−0.111	1.324	−0.359
	(0.804)	(−0.527)	(1.439)	(−1.251)
UNSC election month	−0.0534	−0.234	−0.566	0.105
	(−0.801)	(−0.674)	(-0.724)	(0.304)
UNSC election period (election month until last month of serving on the UNSC)	−0.173	0.0173	−0.0261	0.0543
	(−1.290)	(0.0644)	(−0.0367)	(0.167)

Note: Stars indicate significant p-values (*** p < 0.01, ** p < 0.05, * p < 0.1). For the exchange rate, an increase implies a depreciation vis-à-vis the US dollar.

The second and third rows in Table 2 show the impact of UNSC temporary membership on financial markets. Again, the effects are not statistically significant at conventional levels of confidence. For the UNSC election month itself (second row in Table 2), the effect on bonds is close to zero, while the other variables show conflicting signals: bills slightly decrease in a positive market reaction, while stock price development is negative and the exchange rate depreciates. For the entire election and serving time on the UNSC (third row in Table 2), which is the variable we focus on in the main regression, the market reaction is close to zero - except for bond yields, which slightly decrease. Overall, we conclude that the short-term market reaction to temporary UNSC membership is somewhat inconsistent, but overall, very limited.

Main regression results

In our main regression, we analyze the interaction effect of a new IMF program and contemporaneous membership in the UNSC on key financial market variables. Table 3 offers details on the main regression results, while Figure 5 provides a visualization of the results. The key finding is that a new IMF program is consistently associated with a negative reaction of financial market variables if the country also is a temporary UNSC member (grey bars). This is not the case for new IMF program approvals when a country is not a temporary UNSC member (blue bars). The red arrows in the Figure show the difference made by UNSC temporary membership on the new IMF program’s effect on financial markets, which is also the coefficient of the interaction term shown in the first row of Table 3.

Figure 5. Financial market reaction to a new IMF program depending on temporary membership in the United Nations Security Council (UNSC).

Notes: Grey bars show the coefficients of an interaction term between a dummy for a new IMF program and a dummy on temporary UNSC membership in a given month on yields of bonds and bills, growth in domestic stock prices (in local currency), and growth in the exchange rate to US dollars (if considered floating). Blue bars show the coefficients for the dummy on new IMF program without UNSC temporary membership. Coefficients are based on OLS with robust standard errors. For the exchange rate, an increase implies a depreciation of the local currency vis-à-vis the US dollar. The red arrows show the difference in the effect of an IMF program on financial market variables related to UNSC membership.

Table 3. Main regression results.

	(1) Bond yield	(2) Bill yield	(3) Stock prices (growth rate, local currency)	(4) Exchange rate (growth rate, to US dollar)
New IMF program during UNSC temporary membership (interaction term)	0.375	0.953*	−2.222	0.897
	(1.098)	(1.750)	(-0.817)	(0.786)
New IMF program (dummy)	0.0982	−0.111	1.324	−0.359
	(0.804)	(-0.527)	(1.439)	(-1.251)
UNSC temporary membership (dummy)	−0.0577	−0.0188	−0.0265	0.0399
	(-0.608)	(-0.0883)	(-0.0577)	(0.162)
Inflation	0.0845**	0.114***	0.195	0.456***
	(2.050)	(3.252)	(1.659)	(6.432)
F-test: New IMF program and interaction term (p-value in brackets)	1.53	1.76	1.03	0.89
	(0.22)	(0.18)	(0.36)	(0.42)
F-test: UNSC and interaction term (p-value in brackets)	1.66	0.02	0.35	0.34
	(0.19)	(0.98)	(0.70)	(0.71)
Adjusted R-squared	0.958	0.970	0.361	0.341
Number of Observations	19038	14833	15417	14022
Number of Countries	108	94	64	96
Number of Years	27	27	27	25
Number of Periods	324	324	323	299

Note: Coefficients are based on OLS with robust standard errors. Robust t-statistics in parentheses. Stars indicate significant p-values (*** p < 0.01, ** p < 0.05, * p < 0.1). The variable of interest is the interaction term between a new IMF program and United Nations Security Council (UNSC) membership. For readability, only a selection of controls is shown. Time (month) dummies, country dummies, and year-country interaction terms are omitted from the table. The regression with the exchange rate as dependent variable is limited to cases where the exchange rate regime is considered floating at the point in time. Inflation is monthly changes in the CPI index compared to the previous year. UNSC election period starts with the election month and ends at the end of election period. For the exchange rate, an increase implies a depreciation of the local currency vis-à-vis the US dollar.

For bonds, we find that a new IMF program coincides with an additional increase of about 38 bp if the country is also UNSC member at the time. This is on top of the effect in case the country is not UNSC member. A new IMF program is for non-members linked to an increase of only about 10 bp in the bond yield. Thus, we observe that in presence of geopolitical interests, investors tend to sell a country’s bonds (negative financial market reaction) in reaction to a new IMF program, much more so compared to absent geopolitical interests. However, this result is not statistically significant at conventional confidence levels.

For short-term bills, we find that a new IMF program is linked to an increase of about 84 bp for UNSC members, whereas non-members face a decline of 11 bp. The difference of 95 bp is significant at a 5%-confidence level. Hence, while bill yields decrease slightly in reaction to an IMF program for non-UNSC members (which suggests a catalytic effect of IMF lending), investors actually sell the country’s bills in case of contemporaneous UNSC membership. As with bonds, higher bill yields imply a negative financial market reaction.

For stocks, we find that a new IMF program for UNSC members is linked to a decrease in the stock price growth rate by about 0.9 percentage points. Lower stock prices indicate a negative financial market reaction as investors sell stocks. If the country is not UNSC member, a new IMF program is linked to an increase of roughly 1.3 percentage points in the growth rate of stock prices, suggesting a certain catalytic effect of IMF lending. However, the relation we find is not significant at conventional levels. Given the rather limited data on stock markets, only 64 countries enter this regression, compared to close to 100 countries for the other regressions.

Finally, for the exchange rate, we find that a new IMF program for UNSC members is linked to an increase in the depreciation rate of by about 50 percentage points. Such an increase in the exchange rate to the US dollar (depreciation) implies a negative financial market reaction as investors are selling the domestic currency. For non-UNSC members, a new IMF program is linked to a reduction in the growth rate of by about 35 percentage points. Assuming that a country’s exchange rate would be relatively stable vis-a-vis the US dollar in normal times, it implies that the currency appreciates in case of a new IMF program while not being a temporary UNSC member, which again would suggest a catalytic effect of IMF lending. However, receiving an IMF program while being a UNSC member is associated with a strong depreciation. Again, the link we find is not significant at conventional levels. The limitation of the dataset to cases where the IMF considered the exchange rate regime as floating explains the lower number of observations of this regression compared to the other regressions.

The high volatility of both stock markets and exchange rates, as compared to bond and bills markets, mean that much less of its variation can be explained by our model.

The inflation coefficient largely shows the expected pattern. Higher inflation is in line with the classical Fisher effect and standard exchange rate theories associated with an increase in bond and bill yields, nominal stock market returns and a depreciating exchange rate.

Overall, our results confirm the finding by Chapman et al. (2017) that geopolitical interests in a country that receives a new IMF program is associated with a negative reaction to sovereign bonds. We also find effects for short-term bills, stock markets, and the exchange rate that go in the same direction. Nevertheless, it should also be pointed out that the statistical significance of the results is rather low and the number of observations where an IMF program approval with temporary UNSC membership is rather small. Nevertheless, our findings support the idea that geopolitical interests reduce the IMF’s catalytic effect, which might render IMF programs less effective.

Robustness checks

Section A.2 in the appendix provides detailed results for all robustness check regressions. Overall, the main results are relatively robust to several changes in the regression setup.¹⁶ A first robustness check is to see whether the limiting of extreme values by winsorizing affects the results. To allow more extreme values, we adjust the winsorizing cuts from the 1st and 99^th percentiles to the 0.5^th and 99.5^th percentiles. The results remain comparable to those of the main regression.

A second robustness check is to see whether it makes a difference to limit the sample to exclude LICs (as we do in the main regression) compared to keeping all countries in an unconstrained regression, but extending the interaction term on IMF program and UNSC membership by a third dummy on being a LIC. The variable of interest is then the interaction term in case the dummy on LIC is zero. The results remain close to the main regression.

A third robustness check is to see whether our findings change if we replace the monthly dummies by controls like GDP growth, global stock prices, and yields on US bonds and bills. While the size of the coefficients changes somewhat, the signs remain the same and statistical significance levels remain comparable.

Concluding remarks

In this paper, we have analyzed whether and, if so, to what extent financial markets react negatively to new IMF programs that may be driven more than usual by political considerations. To as exogenously as possible measure geopolitical interests we follow the literature and use temporary UNSC membership. Although this certainly does not capture all cases and the full extent by which IMF programs might be triggered by such political considerations, we still believe it is the best measure at our disposal.

In contrast to the rest of the literature and to better reflect the information contained in financial market variables, the empirical analysis is done on a monthly frequency. By furthermore including as many dummy variables as technically feasible together with otherwise available control variables, we believe that we can get as close as currently possible to identifying causal relationships. At the same time, we have in this way made it difficult to find significant relationships. It is therefore perhaps not surprising that our results are not overly strong when each regression is looked at independently. However, the results are consistent and economically sizeable: independent of whether we use sovereign bond yields, short-term bill yields, domestic stock prices, or the exchange rate countries that receive an IMF program while being temporary member of the UNSC see a deterioration in the assessments of financial markets. In particular, bill yields tend to increase by on average almost 100 basis points as compared to a situation in which a country is not member of the UNSC.

For future research, it would be good to expand the dataset to include more years, and hence more cases of countries who receive an IMF program while being UNSC temporary member. This would allow having an even clearer picture of the financial market reaction in such situations.

Acknowledgements

We thank participants of an SNB Brown Bag Workshop, the 2021 European Public Choice Society and the 2022 meeting of the Verein für Socialpolitik, Stephan Schneider, Michael Siegenthaler, and two anonymous reviewers for feedback on previous versions of this paper. The usual disclaimer applies.

Disclosure statement

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Data availability statement

The underlying data and programs to reproduce the results are available on request from one of the authors.

Additional information

Notes on contributors

Lena Lee Andresen

Lena Lee Andresen is Head of Bilateral Cooperation at the Swiss National Bank (SNB). From 2016 to 2018, she served as Advisor to Executive Director at the International Monetary Fund. She holds a PhD in Economics from ETH Zurich.

Jan-Egbert Sturm

Jan-Egbert Sturm has been Professor of Applied Macroeconomics and Director of KOF Swiss Economic Institute at ETH Zurich, Switzerland, since October 2005. He received his PhD from the University of Groningen (the Netherlands) in 1997.

Notes

1 For further examples, see Stone (2004) and Dreher et al. (2015, pp. 127-129).

2 See for example Dreher et al.  (2015).

3 Gehring and Lang (2020) use monthly data in a related, but different set-up. They focus on how IMF programs affect sovereign creditworthiness in general, but do not specifically examine this in a geopolitical context.

4 See Dreher et al. (2013) for an overview of older research on the topic. Dutta and Williamson (2019) provide a comprehensive discussion of newer studies in this context.

5 Krahnke CITATION Kra20 \n \t \l 2055 (2020) gives a good overview of recent theoretical and empirical literature on the IMF catalytic effect, while Cottarelli and Giannini (2003) cover earlier studies.

6 Bird (1996), Knight and Santaella (1997), as well as Steinwand and Stone (2008) give overviews of earlier research on the subject.

7 Kuziemko and Werker CITATION Kuz06 \n \t \l 2055 (2006) were the first to underline the relevance of UNSC temporary membership as a measure of geopolitical importance in the context of aid. Since then, several studies have used UNSC membership to measure US geopolitical interests in a country.

8 See for example Blustein (1998).

9 Although IMF yearbooks provide program data prior to 1993, to go back further in time, data for our dependent variables is also needed. Exchange rates for a few countries are available, but consistent data for interest rates and stock prices is very limited, creating a very unbalanced panel that is prior to 1993 often of poor quality. Additionally, we have decided not to extend the sample prior to 1993 because it would include the Cold War period during which the behavior of the IMF may have been different (see, e.g. Moser and Sturm, 2011).

10 Section A.1 in the Appendix provides summary statistics of all variables used.

11 See McCauley (2012) for an overview of findings regarding the link between emerging markets assets and global capital flows.

12 Winsorization limits the extreme values in the observed data to a predetermined percentile of that data. We set data below (above) the 1^st (99^th) percentile to the 1^st (99^th) percentile. In this way, the influence of outliers on the regression results is reduced.

13 The results are not sensitive to this dummy taking the value of one during the 24 months when most countries are actually temporary members, or when the effective period is extended by a few more months to account for the fact that it may already be known before the official election that a country will soon be a member. Most of the direct impact of temporary membership in the UNSC is captured by the country-year dummies in our econometric analysis presented in Sections 5 and 6.

14 To check for robustness, we run a regression in which the monthly dummies are replaced by variables on annual GDP data, global stock prices, and returns on US bonds and US bills (see appendix, Section A.2). As discussed in Section 6.3, this does not affect our conclusions.

15 Two alternative approaches to clustering (double clustering by year and country and clustering by year only) produce comparable results. We have chosen clustering by country as this appears most relevant to us and turns out to be the most conservative option in our case, i.e., produces the largest standard errors most of the time.

16 We also considered a robustness check of keeping the sample steady for all four dependent variable regressions, but it did not yield sufficient cases where the interaction term on IMF program and UNSC membership was positive.