Who adjusts? Exchange rate regimes and finance versus labor under IMF programs

Abstract

Who adjusts under International Monetary Fund (IMF) programs and why? In this paper, I demonstrate that labor groups adjust when there is a fixed exchange rate regime and international financial groups are strong enough to defend the peg. In that case, the Fund substitutes currency devaluation with lowering labor costs via labor market reform. Lower wages are used to increase competitiveness and support exports. In other words, the Fund makes labor ‘cheaper’ through its labor market reform when money cannot be made ‘cheaper’ due to strong international financial interests. To test this theory, I use a mixed-method approach. I complement the case comparison of Latvia and Hungary in 2008–two very similar cases except for their exchange rate regime and the influence of international finance on their economy–with a large-N study using a global sample of IMF borrowers over the years 1989 and 2014. The paper shows that international organizations such as the IMF might amplify the voice of the strong (financial interests) while making the ‘weak’ (labor groups) weaker via their conditionality.

1 Introduction

International organizations (IOs) often claim to be impartial institutions that are above domestic power dynamics and interest group competition in member states (Keohane et al., 2009). The International Monetary Fund (IMF) in particular claims that its adjustment programs are purely technical and geared toward addressing economic necessities in borrowing countries in order to prevent a future economic crisis (Best, 2007; Clift, 2018; Clift & Robles, 2021). This paper shows that, contrary to this claim, the design of adjustment programs reflects the domestic power balance between groups and that the question of ‘who adjusts’ under IMF programs is partially defined by this power (im)balance. The IMF is not only not independent from domestic interests and interest groups’ influence but also might reinforce power imbalances by amplifying the voice of already ‘strong’ groups and prioritizing their preferences in adjustment program designs against the interest of the ‘weak’.

In this paper, I particularly demonstrate that there is a substitution between currency depreciation and labor market reform. When international financial groups (i.e. international investors in the country) constitute a significant part of the economy (as a ratio of the GDP); they can leverage their mobility ‘vis-à -vis’ the government and the IMF (Culpepper, 2010; Frieden, 1991, 2015). It is well known that internationally oriented finance prefers certainty and predictability in exchange rates (Broz & Frieden 2001; Frieden, 1991; Frieden et al., 2010; Walter, 2008). In conjunction with the government, the IMF then designs a program catering to international finance’s preferences. In that case, labor market flexibilization measures in the program lower labor costs by directly reducing wages and/or indirectly by decreasing employees’ bargaining power ‘vis-à-vis’ their employers. Lower labor costs make products of the country cheaper in international markets, increase exports, and reduce both imports and general consumption in the domestic economy (IMF, 2010a, 2010b; Walter, 2013), and hence provide the required adjustment to address the balance-of-payments problems. This means that labor groups shoulder the main burden of adjustment due to the fear of sudden disinvestment of financial capital. As a result, we observe that countries with stronger financial interests experience labor market reform under their IMF programs more often than the ones with weaker financial interests. In floating regimes, it is expected that currency depreciation in the lead-up to and during the economic crisis lowers the cost of exports and finances the competitiveness gap of the country without much need for a labor market reform. As we will see later however financial interests can still intervene to maintain exchange rate stability even under floating regimes and push the burden of adjustment onto labor groups if they are strong enough.

I use a mixed methodology to test this theory. I first show that the exchange rate regimes and the strength of financial interests play an important role in the choice for labor market reform beyond the pre-existing regulations in the labor market, by comparing Fund programs in Latvia and in Hungary in 2008. Those two cases were very similar in many respects such as labor market regulation measures, firing costs, unit labor costs, trade union density, overall macroeconomic indicators such as GDP per capita income, type of economic crisis in 2008 (i.e. a banking crisis), and geo-strategic alliances with the U.S. and G-7 countries, but differed in terms of exchange rate regime and the contribution of international financial investment to GDP. Latvia, a fixed regime¹ with financial liabilities amounting to 63 percent of its GDP in 2008 before entering an IMF program received a large number of labor conditions with the motivation of lowering labor costs and financing the spending gap. Hungary, a floating regime that underwent a significant currency depreciation in the lead-up to its IMF program and weaker financial interests, did not receive any labor conditions in the same year.

Secondly, I test the association between exchange rate flexibility, the power of international finance, and labor market reform under an IMF program in a global sample of Fund borrowers over the years 1990 and 2008. I show that countries with less flexible regimes receive more labor conditions (by a simple count of labor conditions in programs) and also more stringent ones (such as prior actions and performance criteria as opposed to less stringent conditions such as structural and indicative benchmarks), controlling for selection into IMF programs, geo-strategic interests, economic factors, organizational capacity of labor groups, and other relevant domestic political factors. Furthermore, I demonstrate that the ‘relative power’ of international financial groups and exchange rate regime flexibility interact in leading up to more extensive labor market reform. At higher levels of financial power, the association between exchange rate flexibility and labor market reforms flattens; the substitution effect disappears and labor market reforms replace currency depreciation. This is because presumably, international financial capital does not only invest in foreign currency but also in home currency and risks being exposed to substantial losses in case of a currency depreciation. Whenever financial interests are strong, they push the burden of adjustment onto labor groups and away from currency depreciation. When they are weaker however we see fewer labor conditions at higher exchange rate flexibility. Then the IMF makes the money cheaper rather than labor cheaper. The results are robust to alternative model specifications and measurement choices as well as the inclusion of additional control variables and country- and year-fixed effects. When financial interests are strong relative to other groups, countries substitute currency depreciation with labor market reform under IMF programs.

The paper builds on four separate streams of literature that developed somewhat independently from one another and aims to further them by underlining the linkages between them and by proposing a coherent theory of the substitution of currency depreciation with labor conditions under IMF programs. First of all, the most important message of the paper is that labor is made ‘cheaper’ when money cannot be made cheaper due to the preference of internationally oriented finance for maintaining currency stability and avoiding depreciation. It shows one of the concrete ways in which interests of financial capital and labor are at odds with one another in the process of globalization (Mosley & Uno, 2007; Rudra, 2002). Secondly, previous studies showed that the Fund is usually friendly to financial interests and banks especially if they are involved in the program design (Dang & Stone, 2021; Gould, 2003, 2006) and not as friendly to labor interests reducing their rights and income (Caraway, 2006; Franklin, 1997; Garuda, 2000; Pastor, 1987; Pion-Berlin, 1983; Reinsberg et al., 2019; Vreeland, 2002; Oberdabernig, 2013). Vreeland (2002), for instance, asks whether the lowered income of labor groups under IMF programs might be intentional rather than an inadvertent consequence of programs and invites further research on this. I show that indeed labor’s income is lowered intentionally to lower labor costs, increase exports, and reduce imports and consumption via lower wages when international finance defends the value of currency. Thirdly, Walter (2008, 2013) discusses how vulnerability profiles, and especially foreign currency-denominated liabilities, might make countries vulnerable to external adjustment (currency depreciation) and make an internal adjustment (monetary tightening and austerity measures) necessary. Following her theory of government receptivity to external vulnerability in driving interests, I suggest to add labor groups as a distinct group in the theory rather than discussing them under the umbrella of general consumers and the electorate. Indeed, labor conditions dismantle collective bargaining institutions and reduce the bargaining power of labor groups; lower the minimum wage and pensions hence reducing the current and prospective income; make firing easier with reduced severance payment and unemployment benefits and hence compound labor market risks; and increase the maximum number of hours one can work on a part-time contract and the maximum duration of a temporary contract. As Caraway et al. (2012) show labor groups would resist those changes if they had sufficient organizational capacity. Finally, building on the studies on exchange rate politics (Broz et al., 2016; Walter, 2008; Frieden, 1994a), I suggest that we should not look at the power of labor groups in isolation in avoiding labor market reform but take into account the power of other interest groups such as financial interests relative to labor groups and investigate when they can push the burden onto one another.

In the rest of the paper, I first discuss the theory behind substituting currency depreciation with labor market flexibilization. Next, I compare Latvia and Hungary in 2008 to show the interaction between the Fund and the respective governments and the justification for extensive labor conditions in Latvia as a substitute for currency devaluation. Then, I present the quantitative evidence in support of the theory. The final section summarizes the argument and concludes with some implications for the study of international organizations, globalization of capital, and rights and income of labor groups.

2 Exchange rate regimes, politics of adjustment, and labor conditions

When and why do labor market flexibilization measures substitute a currency depreciation under IMF programs? This section firstly explains the underlying constraints a government faces when it has a currency peg and undergoes a balance-of-payments crisis. Secondly, it explains the macro-economic logic of reducing labor costs as a solution to balance-of-payments problems under currency constraints. Finally, it delineates the interests and considerations of relevant actors such as the government, the IMF, internationally oriented financial capital, and labor groups and explains how and when groups can push the burden of adjustment onto one another. I argue that labor groups lose in terms of income and rights when international finance is strong and pushes the burden of adjustment onto them.

Countries may want to peg their currency for several well-defined reasons in the literature such as fighting inflation, reducing volatility if they are a small economy, fostering international trade, lowering interest rates, and hence supporting growth (Blomberg et al., 2005; Van Poeck et al., 2007). When governments undergo a balance-of-payments crisis under a fixed exchange rate regime, they face a choice of either defending the currency peg or floating their currency (Leblang, 2003). Since floating during a crisis almost always means depreciation, the choice boils down to between maintaining the peg or depreciating the currency (Frieden, 2015). Under an IMF program, rapid currency depreciation is even more likely, as IMF programs often trigger a government crisis (Dreher & Gassebner, 2012), and a lack of trust in the government is one of the primary reasons for currency depreciation (Bernhard, 2006).

When and why do countries choose to keep the peg and what policy options do they have in case they insist to defend the peg? Since the choice is related to available options, we can begin by looking at the available options in the presence of currency constraints. If a country is determined to keep the currency peg, it can increase interest rates to attract foreign capital, impose capital controls to prevent capital flight out of the country, or implement internal devaluation (Leblang, 2003). Forbes and Klein (2015) show that all those options such as capital controls, interest rate increases, and currency depreciation have adverse macro-economic consequences and that countries have to ‘pick their poison’ among these potentially negative options. Broz et al. (2016) add import tariffs as an additional policy option. If a country implements import tariffs, this would be similar to a currency devaluation. In both cases, consumers can buy less from the outside world, which contributes to correcting the balance-of-payments problems. Under an IMF program, the most preferred option seems to be the internal devaluation via labor market reform (see Fig. 1 below).

Fig. 1

Source: Author’s calculations based on Kentikelenis et al.’s (2016) raw conditionality data set

Number of different types of conditions in IMF programs, 1980 and 2014

Labor market flexibilization almost invariably reduces labor income in the short run. Conditions such as dismantling collective agreements and encouraging firm-level agreements (in other words, more decentralized and individual negotiations at the firm level) result in greater inequality in the wage distribution. While the wages of a minority of workers increase, those of the low-skilled majority decrease (Iversen, 1998; Wallerstein, 1999). Measures such as changes in firing costs and employment protection legislation reduce the bargaining power of employees vis-à-vis employers, which indirectly lowers wages. When it is relatively less costly to fire a worker, employers may be more reluctant to provide higher wages demanded by their employees. Privatizations of state-owned enterprises positively contribute to the government budget but often result in lay-offs of public workers (Caraway et al., 2012). Due to the ‘rule of one price’ (i.e. the rule that prices for similar goods and assets converge in different markets), wages in the private sector also go down. Finally, specific conditions on reducing the public sector wage bill result in the dismissal of public sector workers or cuts in their wages and pension rights (Rickard and Caraway 2019). This increases the competition in the labor market and again leads to an indirect decline in wages due to increased supply.

Reduced wages (‘income policy’ in the IMF’s preferred terminology) decrease consumption and ‘foster real depreciation’ (IMF, 2009). They also ‘improve competitiveness and facilitate external adjustment’ (IMF, 2010a, 2010b). Logically, lower unit labor costs reduce production costs and increase competitiveness. This is a standard way of conceptualizing the trade-off between a reduction in wages and an increase in exports (competitiveness) at the Fund. In Greece, for example, the IMF program specified that ‘Unleashing growth potential requires… ensuring collective bargaining institutions that deliver wages commensurate with productivity’ and decentralized those institutions, which reduced overall wages in the economy (IMF, 2010a, 2010b, p. 7). The Fund explains that ‘Currency devaluation is not an option for Greece because of its membership in the eurozone. This means that unit labor costs can only be improved through improved productivity—which is difficult to engineer in the short run—or through wage adjustments…’ (IMF, 2012). Similarly, in Portugal, labor market reforms were recommended to ‘enhance competitiveness through structural reforms’ (IMF, 2011, p. 13).

Among those labor market reforms were the reduction in overall severance payments, reduction in the maximum duration of unemployment insurance, capping unemployment benefits, promoting of firm-level rather than sectoral-level agreements, and decentralizing the collective bargaining process (IMF, 2011, p. 14). In addition to these measures, the Fund may lower pensions, benefits, and the minimum wage, as in Latvia in 2008 (IMF, 2009, p. 4). The duration of temporary contracts might increase, and benefits attached to part-time contracts in relation to full-time employment might diminish again as in Greece in 2010. All these measures lower labor costs for private sector companies and almost invariably result in lower wages across the economy. This is of course very costly for wage-earners primarily because their current and prospective income declines while insecurity and risks in the market increase.

One important point here is the potential vulnerability of labor groups to currency depreciation if they have foreign currency-denominated liabilities, as explained by Walter (2008, 2013). In such a case, labor groups can be predicted to oppose both currency devaluations and labor conditions and prefer other policy alternatives that might finance the adjustment under a fixed regime such as interest rate increases to attract foreign capital, import tariffs, or perhaps capital controls. Empirical evidence, however, suggests that governments under an IMF program predominantly implement labor conditions instead of other potential alternatives. Between 1980 and 2014, in a global sample of IMF programs, as opposed to 1,510 different labor-related conditions, there were merely 119 conditions to increase interest rates and 20 conditions to increase import tariffs. There were no conditions to implement capital controls or sell reserves. On the contrary, conditionality mainly encouraged governments to establish a floor for their international reserves and to abolish capital controls. Iceland in 2008 and Lesotho in 2011, for instance, received the condition of smoothly lifting capital controls (Kentikelenis et al., 2016). Figure 1 summarizes the number of conditions related to each potential policy option.

When and why do governments then insist to maintain exchange rate stability and shift the burden of adjustment to labor groups? Labor market reforms are painful for labor groups and are highly visible to the electorate– perhaps even more visible than the exchange rate regime itself (Broz et al., 2016). When one is fired or faces a lowered severance payment or unemployment benefit in case they are fired, they are expected to be very sensitive to those immediate changes in their lives. Because of that, perhaps labor market conditions are riskier for governments to implement than currency depreciation. Furthermore, labor groups can stage protests, strikes, and riots and challenge the legitimacy of those flexibilization measures (Metinsoy 2024). Despite those political risks, the substitution of currency depreciation with costly labor conditions under IMF programs can be explained by the relative strength of the internationally oriented financial sectors vis-à-vis labor groups in domestic politics. When internationally oriented sectors are strong, they successfully push to maintain currency stability and avoid significant depreciation or devaluation by substituting it with labor flexibilization.

Internationally oriented financial groups prefer maintaining the currency stability (Frieden, 1991; 1994b; Frieden et al., 2010; Walter, 2008). Less flexible regimes provide certainty in case of pre-committed investment decisions of those groups and serve their interests retrospectively as well as helping them to make prospective decisions (Frieden, 1991). A switch to a more flexible regime might create large losses for them if investment decisions are made under a certain currency regime and in domestic currency. Even in the case of gains due to a depreciated domestic currency (for instance, if the investment was made in a foreign currency and it appreciates in terms of home currency in case of domestic currency depreciation), large-scale uncertainty and expected volatility during a crisis may not be preferable for those groups. Furthermore, the process is likely to be endogenous. The type of finance that really prefers exchange rate stability flows into fixed regimes and makes investment choices accordingly such as investing in the home currency thanks to the stability of the exchange rate regime.

The financial interest groups, different from labor groups, have four potential leverages vis-à-vis governments and the IMF. Firstly, they can leverage their mobility and the disruptive effect they can cause on the economy in case of a sudden withdrawal (Blomberg et al., 2005; Culpepper, 2010; Frieden, 2016). Governments and the IMF fear sudden liquidity problems (IMF, 2009, p. 8). Secondly, if bank loans and deposits are foreign currency denominated, ‘devaluation would create large balance sheet effects’ with skyrocketing liabilities for governments in case of currency depreciation (IMF, 2009, p. 10; Walter, 2013). Eichengreen et al. (2005) call this ‘fear of floating’ in case of ‘original sin’—borrowing in foreign currency when borrowing in one’s own currency in international markets is not possible. Small economies (i.e. countries with smaller GDP) such as Latvia are much more likely to be inflicted by ‘original sin’ compared to countries with a bigger GDP such as Hungary (Hausmann et al., 2001). To be sure though, this is almost as much risk for international financiers as the government due to the potential of non-repayment to international creditors. Thirdly, international finance has greater access both to the government (Tsingou, 2015) and to the IMF (Gould, 2006) and hence possesses a greater lobbying power due to its structural importance for the capitalist economy (Culpepper, 2010). Perhaps more importantly, those international financiers often participate in the bail-out programs contributing to the loan packages (Gould, 2003). They can withhold the credit when it is urgently needed. Finally, they command significant ideational power. Hardiman and Metinsoy (2019) for instance show that internationally oriented finance has substantial influence in shaping the debate on ‘sound’ and ‘acceptable’ policies and crowding out alternative policy options from the public space.

Furthermore, international finance is increasingly interlinked across borders and financial investment instruments potentially compounding their joint vulnerability to non-repayment but also increasing their negotiating power at the table. International financial investment is traditionally considered in three main groups: foreign direct investment, portfolio investment, and commercial bank lending (that often goes under ‘other investment’ in the balance-of-payments sheets). Yet, these three instruments are increasingly interconnected. International liabilities in the banking sector, for instance, are highly critical for the currency regime. Commercial banks mainly ‘rely on borrower’s balance sheet to ensure repayment’ and hence they are highly exposed (Dobson & Hufbauer, 2001, p. 47). Portfolio investment is traditionally more liquid compared to traditional bank loans and foreign direct investment; as they can be easily transferred (Rethel & Hardie, 2017, p. 218). However, commercial bank loans are sometimes connected to bonds through the process of securitization, i.e. pooling loans and selling them in the form of debt securities (Reinert, 2020, p. 339). The increasing need to issue these securities in foreign currencies, i.e., the original sin (Eichengreen et al., 2005), makes a depreciation in the country’s currency (such as by switching to a floating regime) very risky for holders of those bonds and for the banks due to the possibility of non-payment. Finally, foreign direct investment is more strongly tied to the country in the short run due to the often physical component of the investment in the country. However, like the link between portfolio investment and commercial bank lending, the profits from direct investment might be moved to other countries and other investment vehicles exposing international investors to currency depreciation and lack of repayment. Hence, as the liabilities of a country in the financial sector increase; currency depreciation becomes increasingly risky for international financial groups and they use their leverage to defend the currency stability².

Households and voters, in general, are harder to derive interests for, as they often hold conflicting and ambiguous interests. Households are also consumers, who prefer a strong currency as this increases their purchasing power. They also prefer exchange rate stability if they have large foreign currency-denominated debt such as mortgages (Walter, 2013). However, they at the same time in large part rely on labor market institutions and are expected to be sensitive to rising risks and uncertainties and vulnerable to wage decreases in the labor market. It might be safe to assume that this group is too heterogeneous with split and ambiguous interests and preferences that they will not be able to form a unified voice in the negotiations unlike, say, financial interests or trade unions. One caveat here is the election periods, during which governments might be more receptive to general voters’ interests (Broz et al., 2016).

The IMF’s motivation in pushing for labor conditions instead of currency depreciation can be understood in three aspects. Firstly, earlier studies such as Walter’s detailed study on Taiwan, Thailand, Hong Kong, and Korea after the Asian financial crisis demonstrate that the IMF staff are often aware of the domestic power dynamics and sensitivities of interest groups (Walter, 2013). The IMF staff also acknowledge that they take into account the ‘political economy’ of the country in program design to increase ‘program ownership’ which might create space for economically important interests to be represented (Hardiman et al. (2019)). As we will see in the case study discussion on Latvia in the next section, the IMF might concede to protecting financial interests even when it disagrees with a certain policy such as keeping the currency peg. Special interests might be especially powerful in pushing the government, which in turn might make the Fund agree to the ‘second best policy’ in terms of societal welfare (Grossman & Helpman, 1999; Mayer & Mourmouras, 2008). Alternatively, international financial institutions including the IMF might cater towards the ideological commitments of the borrowing governments and help them appease their constituencies such as financial interests (Genovese & Hermida-Rivera, 2022). In addition to special interests capturing the policy process at the international level by using their leverage against both the IMF and the government, previous studies demonstrated that the Fund might be more receptive to financial interests and assign bank-friendly conditions (Gould, 2003). Indeed, the institution sometimes relies on those banks in raising the necessary funds for loans to countries (Copelovitch, 2010). Some IMF programs are seen more as the rescue of the exposed banks than borrowing countries (Dang & Stone, 2021). Finally, the Fund is argued to have a neoliberal agenda that might prioritize capital account openness and mobility (Ban, 2015; Chwieroth, 2007, 2015; Gabor, 2010; Nelson, 2014, 2017; Woods, 2006) and flexible labor markets without much regulation and protection (Caraway, 2006). In other words, the Fund might generally be more receptive to financial interests than labor interests.

The main pillars of the international negotiations between the IMF and the borrowing government then can best be characterized as a mutual understanding that (i) labor will bear the burden of adjustment and the currency stability will be protected if the financial sector is powerful enough and can push for its own interests (ii) the government will have access to liquidity and cater to financial interests if they are strong, and (iii) the IMF will coordinate the bail-out financing, design a program and put its political weight behind the program, and, if necessary, be scapegoated for politically costly reforms. The IMF’s primary role in substituting external adjustment with labor conditions is to act as an external anchor and a scapegoat for the painful political choice of labor market reform that perhaps could not otherwise be (realistically) pushed on the labor groups by the government (Vreeland, 2006)³. In other words, governments, that need an external anchor to implement painful labor conditions that would reduce labor’s income, might go to the Fund (Vreeland, 2002) and implement a policy line via the anchor of the IMF that they otherwise cannot. While doing that, they export the domestic power dynamics to the international arena or, looking from the reverse perspective, they include the IMF in domestic power dynamics on the side of the stronger group.

To sum up, in more flexible exchange rate regimes, theoretically, neither the government nor the Fund has control over the value of the currency. In such regimes, the currency is expected to automatically depreciate in response to a crisis (Bernhard 2006; Blomberg et al., 2005). This would ensure that a country’s products become cheaper in international markets and indirectly contributes to financing the spending gap and correcting the balance of payments. In such cases, the IMF might assign fewer labor conditions compared to less flexible regimes. This holds especially if financial interests are not strong. If they are strong, however, they can still push for labor market flexibilization to achieve adjustment via lowered labor costs while maintaining currency stability. As we will see in the empirical part, countries still receive many labor conditions even under flexible regimes if international finance is strong because of the substitution of currency stability with labor conditions.

In the next section, I compare Latvia and Hungary in 2008 under their respective IMF programs and argue that Latvia’s fixed regime and high international financial exposure played a significant role in receiving stringent labor conditions under its IMF program. Strong financial interests can push for defending a fixed regime and hence push the burden of adjustment onto labor groups.

3 Currency pegs and labor market reform: Latvia and Hungary in 2008

Latvia and Hungary under their respective IMF programs in 2008 provide an excellent comparison for delving into labor market flexibilization under IMF programs. Both countries are former Eastern Bloc members and completed their transitions to the market economy after the end of the Cold War (Stone, 2012). They had similar levels of trade union density by the time they borrowed in 2008 (14.4 percent in Hungary and 15.1 percent in Latvia) (OECD, 2014) and similar GDP per capita income (16,348 U.S. Dollars for Latvia and 15,739 U.S. Dollars for Hungary).

They had similar levels of firing costs and similar labor market regulation measures (Adams et al., 2017). Moreover, both received substantial assistance for their transitions and liberalized their economies under the external influence and aid of the EU and the IMF (Stone, 2012). They experienced similar economic shocks during the 2008 global financial crisis. The crisis that started in the U.S. spilled over to Europe and affected foreign capital inflow to both countries, putting their banking sectors under considerable distress and widening their ‘spending gap’ (IMF, 2008, 2009). They borrowed from the Fund to finance their volatile banking systems and to compensate for the drying up of liquidity in their financial markets.

Latvia and Hungary differed in two crucial respects. While Latvia had a currency peg by the time it borrowed from the Fund in 2008, Hungary had transitioned to a floating exchange rate regime earlier in the same year⁴. This is related to the ‘original sin’ and the ‘fear of floating’ discussed in the previous section. In 2008, Hungary had a bigger GDP and hence economic size (150,374 billion in current US Dollars) than Latvia (35,854 billion in current US Dollars) in the same year. Accordingly, the Latvian banking system had greater international liabilities as a percentage of its GDP (63 percent), while Hungary had fewer such liabilities (27 percent of its GDP). While Latvia received a large number of labor conditions under its IMF program, the Hungarian program did not include any labor conditions (IMF, 2008, 2009). This section discusses the strength of international financial groups in Latvia and how the Fund and the government openly envisaged the labor market conditions as a substitute for floating the currency and avoiding a currency depreciation.

The Latvian government borrowed an exceptionally large amount—1.7 billion Euros (1,200 times its quota)—from the Fund on December 12, 2008. The currency peg of the country played a central role in the program design, as explained and repeatedly underlined by the IMF staff, who negotiated the program. The IMF staff, for example, note that: ‘The authorities’ unequivocal commitment to the exchange rate peg has determined their choice of program strategy’ (IMF, 2009, p. 9). In the same report, they explain that ‘The program’s main objectives are to arrest the immediate liquidity crisis and then to ensure long-term external stability while maintaining the exchange rate peg….’ (IMF, 2009, p. 1).

Furthermore, the staff members explain that the option of switching to a floating regime was discussed with the Latvian authorities. After explaining the advantages of a floating regime such as ‘a quick correction of Latvian exchange rate misalignment’, they nevertheless clarify that ‘a change in the peg is strongly opposed by the Latvian authorities and by the EU institutions, and thus would undermine program ownership’ (IMF, 2009, p. 10). Finance in the Eurozone is highly interconnected with potential contingency, which explains the European institutions’ insistence to preserve the peg (Copelovitch et al., 2016; IMF, 2009; Lutz et al., 2019). Latvian authorities’ insistence is perhaps more puzzling. In an interview reported by Lutz and Kranke (2014, p. 320), an IMF staff member states: ‘Latvia is sticking to that peg… It’s amazing how overriding an objective this is… that they are willing to across the board live miserably for several years to ultimately adopt the euro…’ However, Eurozone membership alone does not explain Latvia’s determination not to devalue. It is true that Latvia was a member of the Exchange Rate Mechanism (ERM) II in 2008–a mechanism in the EU that pegs the non-Eurozone currencies to the Euro at a certain rate and aims to prepare them for the Eurozone membership before the final adoption of Euro. The mechanism, however, allows for a 15 percent margin of fluctuations between two currencies. Latvia voluntarily decided to keep only 1 percent of fluctuation between Lat and Euro rather than devaluing towards the 15 percent margin (Lutz and Kranke 2014). In other words, Latvia could devalue within the margin and still become a member of the Eurozone in 2014 as it was planned.

Perhaps what explains the decision was the great exposure and influence of the Nordic financial coalition in Latvia, which was adamantly against any devaluation of the currency and showed great sensitivity towards it (Lutz and Kranke 2014). The Swedish banks and banks of other Nordic countries (Denmark, Finland, and Norway) jointly owned 60 percent of the banking sector in Latvia before the crisis. Their influence on the political economy of Latvia was such that the Latvian Finance Minister shared the government’s plans to request financial assistance from the IMF with Sweden’s Finance Minister first before going to the IMF (Lutz and Kranke 2010). Furthermore, the Swedish government and particularly Riksbank played an important role in the program design.‘The [IMF] mission cooperated closely with…Swedish Ministry of Finance and Riksbank, and Nordic country governments’ as reported by the IMF (IMF, 2009, p. 1). They also participated in the bail-out package. The Nordic countries pledged 1.8 billion Euros as part of the financial aid package (IMF, 2009, p. 19). In addition to the official channels, private financiers and representatives of the IMF and other international financial institutions also gathered as part of more informal settings such as the ‘Vienna initiative’ (European Bank Coordination Initiative Group) designed to bring European and international public officials and private financiers together (Lutz & Kranke, 2010). International financiers in other words had disproportional access to the policymakers and instrumental/lobbying power compared to labor groups.

International financiers also had asymmetrical power over the Latvian government by holding the much-needed tranche. When the social cost of the program started to surge, Prime Minister, Valdis Dombrovskis, suggested re-aligning the debt of households to the current property prices (which significantly fell after the start of the crisis) (Mabbett & Schelkle, 2015). Dombrovskis argued that ‘Some balance has to be found between the interests of borrowers and the interests of lenders’ suggesting that the international banking sector should share some of the burden.⁵ The idea was met with strong opposition from Nordic financial actors. The newspaper, Guardian, reports that an economist they interviewed at a Scandinavian bank in Riga said: ‘This would be economic suicide. The lending would stop overnight.’⁶ Confirming this, the Swedish Finance Minister threatened to hold up the 1.2 billion Euro tranche to Latvia. Also, the proposal met with criticism from the Governor of the Bank of Latvia, Rimšēvičs. The Central Bank officials accused Prime Minister of creating insecurities about a potential currency devaluation (Lutz and Kranke 2010). This is interesting because Prime Minister unequivocally stated that ‘Devaluation is not an option’ in the same statement proposing the mortgage restructuring⁷. Still, as Lutz and Kranke (2010, p. 25) notes, the Latvian Prime Minister ‘tested the water; now he knew that it was hot enough to burn the incautious.’

The Latvian case also explicitly outlines that a switch to a floating regime and labor market reform was viewed as alternatives to one another and that reduced wages were intended to increase the competitiveness of the country. Fund staff explains that they recommended ‘structural reforms to help address a remaining competitiveness gap and support higher growth and employment through stronger exports in the absence of other policy options’ (IMF, 2010a, 2010b, p. 4). They acknowledge that ‘Depreciation would have boosted exports, allowed lower interest rates, and eased pressures on international reserves’ (IMF, 2010a, 2010b, p. 6). However, since the Latvian authorities were unwilling (or unable) to switch to a floating regime; instead of currency depreciation, the Fund explains that it encouraged wage and product price cuts and envisaged that this would boost exports and start the economic recovery (IMF, 2010a, 2010b, p. 7).

Latvia, as a result, received a program loaded with labor conditionality. The “Committee to Promote Wage Restraint” was formed as one of the first steps of the program. The Committee, in cooperation with social partners and labor experts, advised reducing public wages and monitoring private wages (IMF, 2009, p. 19). The program set an indicative target for the government wage bill at 214 million Lat for the end of March 2009 (a cut of more than one billion Lat compared to December 2008—1,248 million Lat) (IMF, 2010a, 2010b, p. 28). Public sector employment was reduced by 25 percent; 14,000 public sector jobs were lost (Walter, 2013, p. 189). By 2010, there was around a ten percent wage cut in the economy as a whole. The rate was higher—thirty percent for public employees. The cut for the private sector might indeed have been higher and underreported due to the pervasive informal economy (OECD, 2017, p. 24) (the full list of labor conditions for Latvia in 2008 is in Appendix I). The unit labor costs declined in Latvia in the fourth quarter of 2008 to -4.7 under the IMF program, down from 8.4 in the third quarter of the same year (OECD, 2022).

In the meantime, there were several large-scale protests in Latvia due to falling disposable incomes. Around 10,000 protesters gathered in Riga on January 4, 2009 to protest unemployment and tax hikes. Protests later turned into a riot when a group of protesters tried to storm the Parliament. More than 30 people got injured and public and private property was damaged⁸. On January 27, 2009, 3,700 farmers marched to demand state aid for their sector, which resulted in the resignation of the Minister of Agriculture⁹. On April 2, 2009, 12,000 public school teachers protested salary cuts. Around 5,000 students and trade unionists protested the 2010 provisional budget of the government on December 1, 2009.¹⁰ The coalition government became unsustainable and resigned in March 2009. However, these protests did not change the course of the program. Perhaps the threat of holding back the 1.2 billion Euro tranche had more influence than public protests.

Hungary tells a very different story than Latvia in 2008. The country borrowed from the IMF on November 4, 2008—approximately one month before the Latvian stand-by arrangement. Similar to Latvia, the Hungarian financial and banking system experienced intense pressure following the 2008 global financial crisis. The country received 12.3 billion euros from the Fund (1,015 percent of its quota at the IMF) to provide the necessary liquidity for its banking sector and to provide reassurances that the country could meet its debt obligations (IMF, 2008, p. 1). Unlike Latvia, however, Hungary did not have a currency peg at the time of borrowing. The country removed the exchange rate band, abandoned its Euro peg, and switched to a floating exchange rate regime in February 2008, before the onset of the crisis (IMF, 2008, p. 6). The main reason for the policy change was to meet the government’s inflation targets. Unsurprisingly, the Hungarian forint (HUF) depreciated quickly after the start of the global financial crisis later that year and began its free fall vis-à-vis the euro on October 1, 2008—a trend that continued for some time (until around March 2009).

The financial sector and its exposure did not play an equally significant role in Hungary as they did in Latvia. To be sure, the Hungarian banking sector was similarly exposed to foreign-currency denominated loans, and the banking sector was dominated by foreign parent banks; 59.9 percent of the market share was owned by foreign banks in Latvia, and 52 percent of banking system assets were owned by foreign banks in Hungary (IMF, 2008, 2009). However, the net incurrence of liabilities in proportion to GDP was very different. Whereas in Latvia it corresponded to 63 percent of the GDP, it was only 27 percent of Hungary’s GDP. The risks that the financial sector posed, in other words, were much higher in Latvia than in Hungary. While ‘the mission cooperated closely with the staff of the European Commission, ECB, World Bank, EBRD, Swedish Ministry of Finance and Riksbank, and Nordic country governments’ in Latvia (IMF, 2009, p. 1), the IMF mission ‘closely cooperated with the European Commission’ staff but not with international financial institutions or banks in Hungary (IMF, 2008, p. 1).

Hungary’s stand-by arrangement did not have any assigned labor conditions, in contrast to Latvia. The conditionality focused on reducing the government debt. The performance criteria established a ceiling for the central government’s primary balance and increasing international reserves. The indicative target within the program established a ceiling for the government’s total debt stock. Another performance criterion was the non-accumulation of external debt arrears (IMF, 2008, p. 7). Unlike Latvia, the Fund did not assign conditions to cut wages in the public and private sectors, reduce pensions, or make the labor market more flexible. The government promised to maintain nominal wages in the public sector and to cut the additional 13th-month salary and pension for public sector workers at the start of the program (IMF, 2008, p. 3). In the end, however, the government provided an allowance that would compensate public employees for cutting the 13th month’s salary (International Monetary Fund, 2009, p. 6). Moreover, small and medium enterprises received wage subsidies if they maintained or created new jobs financed through EU aid (International Monetary Fund, 2009, p. 7). The unit labor cost increased in Hungary under the IMF program unlike in Latvia: it jumped to 3.4 in the fourth quarter of 2008 from -0.36 in the third quarter of the same year (OECD, 2022).

Hungary and Latvia brilliantly show that the exchange rate regime and international financial exposure play a crucial role in program design. They also show that the relative political-economic weight of international financial interests and the level of exposure in international financial markets affect whether and how many labor conditions a country receives. The next section tests this theory in a global sample of IMF borrowers.

4 Quantitative evidence: Fixed exchange rate regimes and labor conditions in Fund programs

This section tests the association between exchange rate regimes, the power of international financial, and labor conditions in a global sample of IMF borrowers between the years 1990 and 2008 and shares the quantitative empirical evidence in support of the theory. The unit of analysis is country-year.

I proxy labor market reform under IMF programs with labor conditions, for which data come from Kentikelenis et al. (2016)’s IMF conditions data set. In the dataset, they code any condition that would negatively affect the benefits and rights of workers and exclude those that might be beneficial for labor interests (they code those beneficial conditions for labor under redistributive policies). Labor conditions include changes in hiring and firing practices, collective agreements, terms of contracts, minimum wage¹⁰, public sector lay-offs, the public sector wage bill, and pension rights. The number of labor conditions ranges from 0 to 13 with a mean of 1.35 in the sample. Romania in 1999 and 2003 and Gabon in 1996 each received the highest number of labor conditions (13 conditions).

To establish comparability with previous studies, I also weight each condition by its importance and give a higher weight to performance criteria and prior actions (the strictest IMF conditions), and a relatively lower weight to benchmarks. Caraway et al. (2012) provide a robust defense for weighing the conditions, arguing that each condition does not have equal importance in the program. IMF staff monitor performance criteria and prior actions more closely than benchmarks: the program is not canceled or the upcoming tranche is not held back if a benchmark is missed. The same is not true for prior actions and performance criteria (although the Board can make exceptions). I use the burden of adjustment measure outlined in Kentikelenis et al. (2016) and multiply the ‘hard’ conditions of prior actions and performance criteria by ‘2’ and add the ‘soft’ conditions (i.e. benchmarks) to the weighed measure of hard conditions (the variable ‘BA3LAB’) (descriptive statistics for all variables are in Appendix II).

Finally, I construct a relative frequency measure for labor conditions by dividing the number of labor conditions by the total number of IMF conditions that year (e.g. fiscal, structural, financial, and redistributive and social policy). Bigger programs with high adjustment targets would include both a higher number of labor and total conditions compared to smaller programs. I demonstrate that exchange rate regime inflexibility increases not only the simple and weighted counts of labor conditions but also the relative frequency of labor conditions within the total number of conditions (see Fig. 2).

Fig. 2

Source: Data on exchange rate regimes come from Ilzetzki et al. (2017)’s Exchange Rate Regimes data binary measure; data on labor conditions come from Kentikelenis et al. (2016)’s IMF Conditionality Dataset

(Weighted) Mean number of labor conditions and relative frequency of labor conditions in floating versus fixed regimes, 1980–2014.

One of the main independent variables in the analysis is the exchange rate regime in the borrowing country. I use Ilzetzki et al. (2017)’s data set on exchange rate regimes, which is an expansion of the earlier Reinhart and Rogoff (2004)’s dataset. The data range from ‘1’ (strict currency pegs or currency union membership) to ‘14’ (freely falling). Regimes are ranked in an order of increased flexibility (Guisinger & Singer, 2010). In addition to this ‘fine’, continuous measurement, Ilzetzki et al. (2017) also provide a 5-point, ‘coarse’ measurement. Once again, for this coarse measure as well, exchange rate regimes are ordered from the least flexible ones (1) to most flexible ones (5). Following earlier studies, I drop the cases where data is missing due to the parallel functioning market (category 15 on the fine measurement and category 6 on the coarse measurement) (Copelovitch & Pevehouse, 2013).

For robustness checks, I draw a data set based on IMF’s Annual Reports on Exchange Arrangements and Exchange Restrictions (AREAER), which covers the years between 1988 and 2013 (IMF, 2020). This is a binary measure. I code ‘1’ if the country has any type of fixed regime in those reports, and ‘0’ otherwise. In 2013, according to the IMF’s reports, the most commonly preferred exchange rate regime in the world was the conventional peg (42 countries out of 168), while floating regimes were the close second (37 countries) (IMF, 2020).

Figure 2 shows that fixed regimes receive on average a greater number of labor conditions and a higher number of quantitative performance criteria and structural performance criteria compared to floating regimes. Finally, labor conditions constitute a bigger part of the total number of conditions for fixed regimes (i.e. they have a higher relative frequency in fixed regimes compared to floating ones). Overall, fixed regimes receive approximately 0.5 more conditions compared to floating ones. Approximately 76 percent of countries in the sample that received at least one labor condition between the years 1980 and 2014 had a fixed exchange rate regime in that year. A simple t-test demonstrates that the difference between the mean numbers of labor conditions for fixed and floating regimes is statistically meaningful and fixed regimes receive more conditions on average (p = 0.0001).

The theory in this paper posits that when internationally oriented finance is strong in a country, the country will maintain its fixed regime or avoid currency depreciation under a floating regime and hence will receive more labor conditions to compensate. To measure the significance of international finance for an economy, I look at ‘investment income debit’ (liabilities that are created by residents of the rest of the world by investing in the country) in the balance-of-payments sheet. The investment income debit is the sum of three types of investment: (i) foreign direct investment, (ii) portfolio investment, and (iii) ‘other investment’ (which is mainly commercial bank lending (Reinert, 2020)). I turn the total investment income debit into a ratio by dividing it by GDP. The investment income measure is ideal for this study as it looks at the yearly income generated by investment inflows rather than the stock of assets and liabilities in a country. It is one-year lagged to isolate the impact of the IMF program on investment flows (Chapman et al., 2017).

The ‘vulnerability to external finance’ measure can be written as follows:

$${{\text{Vulnerability}}}_{i,t}=\frac{{\mathrm{Investment\;income\;debit}}_{i,t}}{{{\text{GDP}}}_{i,t}}$$

In the measure, i stands for the country; t stands for the year. Data come from the IMF Balance of Payments dataset Version 6 (BPM6)¹¹. One can argue that internationally oriented domestic finance can also pressure the government and the IMF to maintain the currency stability. Since IMF program countries are often receivers of international financial investment with scarce capital outflows, this impact is expected not to be very strong. For robustness checks, however, I also look at the total sum of investment income debit and investment income credit as a ratio of GDP.

As predicted by earlier literature, countries with a higher investment inflow as a ratio of GDP are significantly more likely to keep a fixed regime and less likely to switch to a floating regime controlling for GDP, GDP per capita income, volume of trade, the regime type (democracy) and panel fixed effects (p = 0.003, n = 3,524 country-year observations). Furthermore, fixed regimes are more likely to attract foreign financial investment once again controlling for GDP, GDP per capita income, the volume of trade as a ratio of GDP, regime type, and panel fixed effects (p = 0.048, n = 2,285)¹².

In addition, the inflow of international finance does not seem to have a direct association with labor conditions that is unmediated by exchange rate flexibility increases. Figure 3 shows that there is not a discernible difference between countries with lower international inflows and higher international inflows in terms of the number, strictness, or the relative ratio of labor conditions within total conditions. As we will see in the analysis section, the influence of vulnerability to international finance has the strongest impact when there is exchange rate flexibility. As explained by the theory, financial capital is attracted to avoiding exchange rate risks. In fixed regimes, labor conditions substitute switching to floating regimes or devaluation. In more flexible regimes, international finance intervenes to push the burden of adjustment onto labor groups if they are strong. Next, I test the theory in a global sample with relevant control variables for confounding effects.

Fig. 3

Source: Data on ratio of international debit income to GDP come from IMF Balance of Payments dataset Version 6 (BPM6); data on labor conditions come from Kentikelenis et al. (2016)’s IMF Conditionality Dataset

(Weighted) Mean number of labor conditions and relative frequency of labor. Conditions in Countries with Lower versus Higher Financial Power Regimes, 1980–2014.

4.1 Model building

It is well documented in the literature that IMF borrowers share some systematic commonalities such as an unfolding economic crisis, ideological proximity to the US, and a democratic regime (Copelovitch, 2010; Stone, 2008; Vreeland, 2003, 2006). Perhaps more importantly for this paper, countries with a fixed exchange rate regime might avoid borrowing from the Fund (Trudel, 2005). Similarly, countries with a higher ‘vulnerability’ to international financial investment and a more regulated labor market might be more reluctant to go to the Fund. They might anticipate painful, politically costly measures. Alternatively, they might be more likely to borrow and receive labor conditions as these reforms might be too costly to be implemented without an external anchor and with significant opposition from domestic groups (Vreeland, 2006). In other words, factors that make countries more (or less) likely to go to the IMF might also affect whether or not they receive labor conditions.

To account for this endogeneity, I instrument selection into IMF programs following Lang (2021) by an interaction term of IMF’s liquidity and the average participation of a given country in IMF programs. As Lang (2021) carefully demonstrates, when the IMF’s resources are stretched in a particular year, repeat borrowers—those with a higher average participation score—are more likely to be prioritized in lending arrangements. Hence the measure has a high predictive capacity for selection into IMF programs. Also, by interacting a time-variant measure, i.e. the IMF’s liquidity, with a time-invariant measure, i.e. a country’s average IMF program participation, we get an excludable instrument. As Lang (2021) explains a third country borrowing from the Fund theoretically would not have an immediate impact on another country’s likelihood of borrowing from the Fund. Data on IMF’s liquidity come from Lang (2021) and on IMF program participation from Dreher et al. (2009).

In the model building, I follow Stubbs et al. (2020)’s recommended approach of running an instrumental variable analysis over simultaneous equations to address the concerns of endogeneity. In studying the impact of IMF programs, they recommend using three simultaneous equations: one for selection into the IMF, one for selection into a particular branch of conditionality, and one for the impact of conditionality on political or economic outcomes. Here, I fit only the first two stages: selection into IMF programs on the basis of variables that also affect selection into labor conditionality. The simultaneous equation here means that selection into IMF and selection into conditionality are mutually determined based on the same variables¹³. I fit the model using Maximum Likelihood Estimation (MLE). The model also includes time-fixed and country-fixed effects in order to control heterogeneous country and time effects.

4.2 Control Variables

In the analysis, I control for several political and economic variables that were shown to influence IMF’s conditions in general and labor conditions in particular. Caraway et al. (2012) previously demonstrated that the IMF would assign fewer conditions in democratic countries where labor groups have the potential to disrupt the government. They measure labor power by looking at the ratio of skilled labor to unskilled labor multiplied by the inverse ratio of surplus labor using Rudra (2002)’s potential labor power (PLP) data set. In this data set, highly-skilled labor is operationalized as the number of employees in certain manufacturing sectors that require greater specialization, and low-skilled labor is measured as the number of employees in more generic-skill-based manufacturing sectors (the list of high-skilled and low-skilled manufacturing sectors is in Appendix VI). The dataset is available for the years between 1980 and 2000. To extend it beyond 2000, I code the ratio of high-skilled to low-skilled labor for the years between 2000 and 2014 based on the INDSTAT 2 ISIC Revision 3 data set from UNIDO (2018). Data on surplus labor between 2000 and 2014 come from Donno and Rudra (2019). Following the same approach, I take the ratio of high-skilled to low-skilled labor and multiply it with the inverse ratio of surplus labor. Following Caraway et al. (2012), I measure democracies on a 20-point scale, in which ‘0’ denotes the most authoritarian countries and ‘20’ the most democratic ones. Data on democracies come from the Polity II project. In the analysis, I look at the interaction effect between the PLP and the democracy variables.

Regulated labor market

The IMF might assign a higher number of, and more stringent, conditions in countries where employment is heavily protected; where there are safeguards against overtime work; collective agreements are extended in the labor market; and where there are stringent conditions for dismissal (Blanchard et al., 2014; Caraway et al., 2012). To control for this impact, I create a regulated labor market variable as a composite variable, which includes indicators for firing costs, collective agreements, and wage protection, and hence extend to the areas covered by IMF labor conditionality. The data come from CBR Labor Regulation Index (Adams et al., 2017). (The full list of variables included in the composite index is in Appendix III). Higher numbers indicate a more regulated labor market and greater rights for workers. The measure ranges from 0.5 (minimum regulation) to 7.95 (maximum regulation). It is one year lagged in order to exclude the IMF program’s impact and the changes it brings to the labor market in a particular year.

US ally

Previous studies have shown that the allies of the U.S. receive fewer conditions due to the specific weight of the U.S. in the Fund’s decision-making (Dreher, 2006; Dreher & Gassebner, 2012; Dreher & Jensen, 2007; Dreher et al., 2015; Stone, 2004, 2008). Furthermore, Lipscy and Lee (2019) show that the allies of the US and G-7 countries might have fewer financial regulations and a higher likelihood of a crisis since they know they will be bailed out by the IMF if they run into problems. I add the UN General Assembly voting in line with the U.S. (US ally variable) to the analysis. Higher values indicate greater alignment in voting trends and hence a closer alliance between the borrowing government and the U.S. For robustness checks, I also look at the collective impact of alliance with G-7 countries measured as voting with the G-7 countries in UNGA (G-7 ally) and being a UNSC temporary member (UNSC member) measured as a binary variable, ‘1’ for temporary members and ‘0’ otherwise. Data come from Dreher and Gassebner (2012) for General Assembly voting and from Dreher et al. (2009) for UNSC membership.

Elections

Broz et al. (2016) show that elections play a critical role in the adjustment strategies of countries in case of balance-of-payments problems. Governments might postpone visible adjustment measures such as implementing labor conditions until after the elections. Furthermore, they might avoid borrowing from the IMF altogether in election years (Rickard & Caraway, 2014). I code the election variable ‘1’ if there is a parliamentary or an executive election in a particular year and ‘0’ otherwise. Data come from the Database of Political Institutions (DPI) dataset.

Economic crisis

Finally, countries with a declining GDP per capita and GDP income might be more likely to borrow from the Fund and receive labor conditions for addressing the economic crisis. More importantly for this paper, those with declining foreign exchange reserves might request IMF assistance especially if they are keen to defend the peg. Data for variables on GDP, GDP per capita, and foreign exchange reserves (in US Dollars as a percentage of GDP) come from the World Development Indicators (WDI) dataset of the World Bank.

4.3 Empirical results

Table 1 reports the results of the instrumental variable analysis over two simultaneous equations. It shows that the interaction between exchange rate flexibility and the vulnerability to international finance measures is statistically highly significant in predicting both measurements of labor conditions, i.e. stringency of labor conditions and the simple count of labor conditions.

Table 1 Exchange rate flexibility, vulnerability to international finance, and labor conditions

The marginal impact of the interaction term is substantively small when all other variables are held at their means, but it might make an important difference in the context of labor conditions. The model, where the simple labor conditions count is the dependent variable, provides the most intuitive substantive interpretation. The mean number of labor conditions for countries under IMF programs in the sample is 1.3 and the standard deviation is 2.2. In a fixed regime (a pre-announced peg or currency board -’2’ on the 14-point exchange rate flexibility scale), vulnerability to international finance results in 1.8 additional labor conditions at its maximum and an 0.5 additional labor condition at its mean. An additional labor condition might mean a significant reduction in the minimum wage, lay-off of a substantial number of public sector workers, or a significant cut in the collective bargaining coverage. In other words, ‘real life’ associations are even greater than the mathematical expressions shared here.

Figure 4 illustrates the marginal change in the strictness of labor conditions (weighted labor conditions) predictions at different levels of exchange rate flexibility. It shows that as flexibility of the exchange rate regime increases and moves towards freely floating regimes (13) or freely falling ones (14), the number of (strict) labor conditions decline at all levels of vulnerability to international finance (investment income debit as a ratio of GDP). When the exchange rate regime is more flexible, we expect to see fewer (strict) labor conditions in general.

Fig. 4

Marginal effect of vulnerability to international finance at different levels of exchange rate flexibility. Notes: Marginal impact of the’vulnerability’ variable on the strictness of labor conditions at the first three different quartiles of the ‘vulnerability’ variable. Outer lines indicate the 95% confidence interval for the predicted marginal effect. Marginal impacts are calculated based on Table 1 Column 1 results

Figure 4 also shows that as the vulnerability to international financial flows increases, the likelihood of receiving labor conditions also increases. For example, when there is a de facto crawling peg (‘7’ on the exchange rate scale), the number of predicted labor conditions is 0.4 in case of minimal investment income debit, 0.5 at the median of investment income debit, and 0.6 when it is at its 3rd quartile (higher). Finally, and perhaps, more interestingly, Fig. 4 shows that as the vulnerability to international finance increases, the marginal impact curve becomes flatter. This might be explained by how much voice the international finance has and how able it is to push the burden of adjustment to labor groups. When it constitutes a smaller part of the GDP of a country, it may not have as much say. As its importance increases, however, it might pull its political weight to contain the decline in the currency value even at more flexible exchange rate regimes and push for achieving adjustment via labor conditions. Logically, if labor costs go down, some exports growth might be achieved via the lowered production costs. This would ensure foreign exchange flowing in to the country and hence containing the decline in the value of the country’s currency.

In line with Hunermund and Louw (2023)’s convincing argument, I do not interpret the coefficients of the control variables, since they are correlated with other unobserved factors. One important aspect to note, however, is that once international financial investments are accounted in the model, exchange rate flexibility loses its statistical significance even though it has a very strong binary association with labor conditions (see Fig. 2). Both variables–exchange rate flexibility and vulnerability to international finance–are significant at 10 percent level in the multivariate analysis when the other is excluded. Following the theory, we can explain this with the fact that when international finance is weaker adjustment in flexible regimes come from currency depreciation as opposed to labor flexibilization. In the next section, I check for the robustness of these results.

4.4 Robustness checks

In this section, I fit three alternative statistical models to check for robustness of findings: a negative binomial model with robust standard errors clustered across countries and with the country- and year-fixed effects where the dependent variable is the simple count of labor conditions, an OLS with robust standard errors clustered across countries where the dependent variable is the weighted labor conditionality and an OLS with fixed effects where the dependent variable is the relative frequency of labor conditions within total conditions. Particularly, negative binomial regression models are the power horse models in the literature when the dependent variable is a count variable such as labor conditions and is non-normally distributed (Cameron & Trivedi, 2013; Caraway et al., 2012; Robertson & Teitelbaum, 2011) (the distribution of the labor conditions variable is in Appendix IV). All models control for factors that are previously shown to affect IMF’s conditionality and the country- and year-fixed effects model control for various idiosyncratic political and economic factors within a country and a particular year. Table 2 shares the results.

Table 2 Robustness checks

Table 2 shows that the interaction between the vulnerability to international finance measure and the exchange rate flexibility remains statistically significant with these alternative model specifications as well. These models should not, however, be over-interpreted, as they do not account for endogenous selection into IMF programs and into labor conditionality. They provide, however, additional empirical support that the association between the vulnerability to international finance and exchange rate flexibility variables does not solely stem from modeling choices.

In addition, I re-run the models by replacing the composite ‘regulated labor market’ measure with firing costs alone following Caraway et al. (2012); including an alliance with G-7 countries instead of an alliance with the US, and adding the UNSC membership status to the analysis.

With these alternative measurement choices as well, the association between vulnerability to international finance and exchange rate flexibility and labor market reforms under IMF programs remains highly statistically significant (the results are in Appendix V). I also control for currency union membership, particularly Eurozone and CFA Franc zone member countries and, with this control variable added as well, the association remains robust.¹⁴ There is compelling empirical evidence that there might be a substitution between currency depreciation and labor market flexibilization and this relationship is associated with the strength of financial interests.

5 Conclusion: Increasing global capital mobility and labor interests

The origin of the IMF is rooted in exchange rate stability, and this paper shows that exchange rates still influence IMF’s conditionality. Particularly, the Fund assigns labor conditions when the currency cannot be depreciated due to high vulnerability to international financial investment. When internationally oriented financial groups are strong and constitute a significant part of the economy, governments and the IMF cater to their need for currency stability and predictability. They implement extensive labor conditions to lower labor costs. Lower wages reduce labor costs, make production cheaper, make the economy more competitive in international markets, lower consumption in the domestic economy, and thus finance the spending gap. This of course puts the burden of adjustment disproportionately on labor groups. Furthermore, it means that the IMF amplifies the voice of international financial interests at the negotiation table and acts as a scapegoat facilitating the implementation of politically costly labor conditions in domestic politics. It can be argued that the Fund becomes part of domestic power competition in those cases on the side of the ‘stronger’ party via its conditionality.

This study contributes to the existing body of knowledge on international organizations in several ways. There is a long line of excellent studies documenting the adverse impact of IMF programs on labor groups. This study unpacks the income redistribution away from the labor groups under Fund programs and demonstrates a concrete mechanism, i.e. the substitution of currency depreciation with labor market flexibilization, that leads to lowered income and rights for labor groups. Furthermore, it shows an additional dimension in the clash of international finance and labor in the process of globalization especially with the increasing mobility of global capital and how IOs might amplify the voice and representation of financial interests.

Furthermore, a highly mobile international capital seems to increasingly force (especially small) countries such as Latvia to a particular corner of the ‘impossible trinity’, i.e. to a fixed regime combined with capital mobility removing the choice of monetary policy autonomy altogether. Largely immobile labor groups, that are tied to the country due to various linguistic, cultural, personal, and migration barriers, have increasingly lower leverage vis-à-vis governments and international organizations. If election periods become the only time power-holders are remotely responsive to labor interests, this power imbalance between finance and labor groups means strong damage to democratic representation. By outlining how labor market reform and lower wages and protection are used as a substitute for currency stability, I hope that this study deepens the conversation on the role of governments and IOs in ensuring fair and equitable representation and outcomes for all groups in global politics.

Building on this paper, future studies can unpack the origins of conditionality further and look at which types of conditions come from the IMF, the government, and the interest groups. This study largely drew on the case comparison deriving actors’ motivations and preferences. Building on this study as well as on Mayer and Mourmouras (2008); Grossman and Helpman (1999); Alesina and Drazen (1989), future studies can clarify strategic interactions of actors in a formal model. Such a study would immensely contribute to the literature clarifying when scapegoating is at play, when special interest groups dictate their preferences, and when the IMF’s preferences (or the way the IMF ‘sees’) take precedence.

Furthermore, future studies can incorporate the main insight of this paper and can look at the relative strengths of groups vis-a-vis one another rather than their absolute strengths in influencing different branches of conditionality. In this paper, I argued that financial interests can dominate both government’s and the IMF’s preferences at times. But, the financial interests are a particularly strong group. Looking at other issue areas and branches of conditionality might yield different insights and further our knowledge of how relative strength of different groups influence conditionality.

Future studies can also unpack the role of the government ideology in receptivity to financial interests. In this study, due to data limitations, I could not unpack this impact. Drawing on Genovese and Hermida-Rivera (2022) and this study, one can look at whether left-wing governments are less likely to implement labor market reform catering to the interests of the finance industry. Alternatively, one can speculate that left-wing governments are more likely to be met with suspicion by financial markets and might be in greater need to be ‘on the good side of the markets’ (Gunaydin, 2018; Metinsoy, 2022). Future studies can look at which of these mechanisms is at play.

Appendices

Appendix I: List of labor conditions in Latvia under its IMF program in 2008

• The government will sign a protocol that stipulates a 15 percent reduction in local government employees’ compensation (Prior action)

• An indicative ceiling on the general government wage bill. (Quantitative indicative target)

• National Tripartite Co-operation Council to establish a Committee to Promote Wage Restraint. (Structural benchmark)

• Wages: reform the Committee to Promote Wage Restraint by involving the social partners and outside labor market experts (Structural benchmark)

• Wages: prepare a comprehensive report on proposed revisions to the public-sector wage grid and the relative wage adjustment across public institutions. (Structural benchmark)

• Wage bill (Indicative benchmark)

• Pension reform (structural benchmark)

• Put in place a wage-setting mechanism in line with the fixed exchange rate regime. (Soft condition)

• Prepare an active labor market policy (ALMP) strategy that will replace the WWS program (structural benchmark)

*There were no labor conditions in the IMF program of Hungary.

Appendix II

Table 3.

Table 3 Summary statistics for variables in quantitative analysis

Appendix III: Indicators included in the regulated labor market variable

• The maximum duration of fixed-term contracts

• Overtime premia

• Limits to overtime working

• Maximum daily working time

• Legally mandated notice period

• Legally mandated redundancy compensation

• Minimum qualifying period of service for the normal case of unjust dismissal

• Law imposes substantive constraints on dismissal

• Extension of collective agreements

• Lockouts (Equals 1 if lockouts are not permitted. Equals 0 if they are.)

Appendix IV: Distribution of labor conditions variable

Fig. 5.

Fig. 5

Distribution of labor conditions variable

Appendix V: Additional robustness checks

Table 4.

Table 4 Robustness checks with alternative measurements

Appendix VI: List of high-skilled and low-skilled sectors

Low-skilled industries: (17) Textiles, (18) wearing apparel and fur; (19) leather, leather products, and footwear; (20) wood products (excl. furniture); (21) paper and paper products; (25) rubber and plastic products; (26) non-metallic mineral products; (28) fabricated metal products; and (36) furniture, manufacturing n.e.c.

High-skilled industries: (22) printing and publishing; (23) Coke, refined petroleum products, nuclear fuel; (24) Chemicals and chemical products; (28) Fabricated metal products; (29) Machinery and equipment n.e.c.; (30) Office, accounting and computing machinery; (31) Electrical machinery and apparatus; (32) Radio, television and communication equipment; (33) Medical, precision and optical instruments; (34) Motor vehicles, trailers, semi-trailers; (35) Other transport equipment.

Appendix VII: Controlling for currency union members

Table 5.

Table 5 Exchange rate flexibility, vulnerability to international finance, and labor conditions