“Selective Friendship at the Fund”: United States Allies, Labor Conditions, and the International Monetary Fund’s Legitimacy

This article discusses the International Monetary Fund’s recent effort to garner legitimacy by incorporating the reduction of economic inequality in its lending programs. It argues that the impact of the US as a major shareholder on condition‐ ality and geopolitical considerations beyond objective and measurable economic necessities detract from these efforts to expand legitimacy. Using a panel data analysis of International Monetary Fund programs between 1980 and 2014, the article shows that US‐allied left‐wing governments receive a larger number of labor conditions in their programs compared to non‐allied and right‐wing governments. The article argues that this is part of left‐wing governments’ strategy of main‐ taining their alliance with the US and demonstrating ideological proximity. In exchange, the US uses its influence to secure fewer conditions in total for its allied governments. This not only shifts the burden of adjustment on labor groups but also harms the Fund’s procedural legitimacy, as conditions are not objectively determined. It also has adverse implications for outcome legitimacy by distorting economic policies and outcomes and increasing income inequality.


Introduction
The International Monetary Fund (IMF) is one of the most controversial international organizations in the global governance system. Scholars often discuss the IMF's "legitimacy crisis" (Best, 2007;Clift & Robles, 2021;Seabrooke, 2007) and the institution's continuous efforts to reform and address these challenges (Metinsoy, 2019;Momani & Hibben, 2017;Woods, 2006). As part of a recent effort to address rising criticisms and garner legitimacy, the Fund incorporated tackling economic inequality in its lending programs as a "macro-critical issue," i.e., an issue that is critical to macroeconomic stability (Ostry et al., 2016). Scholars agree that this effort has been limited in its success (Best, 2007;Clift & Robles, 2021;Gronau & Schmidtke, 2016;Nunn & White, 2016). They explain this by the Fund's narrowly focused expertise on "classic economics" and its lack of exper-tise in the field of politics (Best, 2007), the bureaucratic structure of the Fund, and, relatedly, the institutional and ideational setbacks for change (Clift & Robles, 2021;Nunn & White, 2016), as well as fragmented and slow-paced change in Fund policies (Kaya & Reay, 2019). Another group of scholars look at socially, economically, and politically harmful consequences of Fund lending programs such as exacerbating poverty, instilling political instability, and harming economic growth (Abouharb & Cingranelli, 2007;Hartzell et al., 2010;Lang, 2021;Oberdabernig, 2013;Przeworski & Vreeland, 2000), which indirectly diminishes the legitimacy of the Fund's lending programs. Finally, scholars have argued that the influence and the privileged role of powerful states in terms of voting rights at the Fund and less voice for developing countries in comparison harm the Fund's legitimacy (Guastaferro & Moschella, 2012;Seabrooke, 2007;Woods, 2006), In this article, I propose an alternative argument based on geopolitical interests as to why the IMF cannot overcome its legitimacy gap and cannot reduce income inequality in borrowing countries. I argue that US-allied left-wing governments receive more labor conditions compared to non-US allied and right-wing and centrist governments. The argument builds on the US influence on the design of conditionality and its role in securing favorable "deals" for its allies. Scholars have previously considered how an alliance with the US translates into larger loans and fewer conditions for borrowing countries (Copelovitch, 2010;Dreher & Jensen, 2007;Stone, 2008). In this article, I look at how an alliance with the US can result in a higher number of and more stringent labor conditions for left-wing governments. I argue that labor market reform under an IMF program helps these governments preserve their geostrategic alliance with the US and signal their pro-market disposition to international financial markets. In Eastern European borrowers, in addition to this, an alliance with the US coincides with signaling a break with the Soviet past. In exchange, these governments reinforce their alliance with the US and demonstrate that they do not have a "radical" leftist agenda. The alliance may then serve allied governments (in the narrow sense of the government as an institution), for instance, by receiving more aid from the US or general support for their diplomacy and foreign policy goals as well as helping them bypass domestic opposition by shifting the blame onto the IMF. This support, however, comes at the expense of domestic labor groups. In other words, the article brings in a qualification for the earlier studies that an alliance with the US results in more lenient treatment at the Fund: It shows that an alliance with the US does not necessarily result in more favorable outcomes for all domestic groups. This "selective friendship" and shifting the burden onto labor groups not only violates the Fund's claim of assigning conditions based on objective and measurable economic indicators but also makes it serve as a "backdoor" for geostrategic alliances and goals. This is highly damaging for an international organization that claims legitimacy based on its reputation as a technocratic institution.
Finally, I argue that labor conditions in Fund programs are intimately linked to rising inequality in lending programs. As discussed in a recent IMF staff discussion article, labor market flexibility has a considerable influence on increasing income inequality (Dabla-Norris et al., 2015). IMF labor conditions that dismantle income-and job-protection measures in the labor market and bring greater flexibility cause salary increases for a minority of workers, whereas salaries of the majority decrease with the declining security and regulation. Since the majority of the wider public in borrowing countries are wage earners, income inequality substantially increases. This contradicts the Fund's declared policy goals of reducing income inequality in borrowing countries and significantly compromises its outcome legitimacy (i.e., the outcomes of the programs).
In a panel data analysis of the years between 1980 and 2013, I show that left-wing governments allied with the US receive more labor conditions controlling for selection into IMF programs, time trends, labor market indicators such as the level of regulation and flexibility in the labor market, and political indicators such as democratic regimes and elections in a particular year. I look at three separate indicators of labor conditions: total number of labor conditions in a lending program, relative frequency of labor conditions among total conditions (i.e., number of labor conditions divided by the total number of conditions), and the stringency of conditions such as prior actions and performance criteria, fulfilment of which determine whether the country can receive a loan from the Fund. With all three measurements, left-wing governments allied with the US receive a larger number of labor conditions than non-allied and right-wing governments. This shows that governments can pursue their own self-interest using their alliance with the US and the IMF and can shift the burden downward onto society. This, undoubtedly, has a significant adverse impact on the legitimacy of the Fund in the eyes of the wider public.
The rest of the article is organized as follows: The next section offers a survey of the existing studies and points to gaps in and the contribution of this study to the extant literature. Section 3 offers a more detailed theorization of how the alliance with the US translates into more labor conditions despite fewer total conditions in programs and how this affects the Fund's legitimacy. Section 4 provides the empirical evidence for the association between an alliance between the US and the borrowing government, the left-wing ideology of the government, and a greater number of labor conditions in lending programs. The final section summarizes the argument.

The International Monetary Fund and Its Legitimacy Crisis
The IMF's continuous "legitimacy crisis" has previously been discussed under three subheadings in the literature with corresponding advice on how to address it. Firstly, the US's privileged role in terms of voting rights at the Fund and G7 countries making up more than 60% of all votes are identified as problematic for the remaining IMF members. Considering that G7 members almost never borrow from the Fund and there are repeat borrowers among developing countries, the lopsided voting structure has been heavily criticized (Guastaferro & Moschella, 2012;Seabrooke, 2007;Woods, 2006). In response, the corresponding legitimation attempt by the IMF consists of revising quotas and giving developing countries a greater voice (Gronau & Schmidtke, 2016). The Fund has, however, gone through multiple rounds of revisions to its voting system without a corresponding increase in its legitimacy (Metinsoy, 2019).
A second factor is the narrowly defined and applied economics expertise at the Fund, usually following the teachings of "classical economics," which may not correspond to the political realities in borrowing countries (Best, 2007;Clift & Robles, 2021;Nunn & White, 2016). Furthermore, the Fund is known to hire from a narrow pool of macroeconomists usually trained in Anglo-Saxon countries (Chwieroth, 2015;Nelson, 2014Nelson, , 2017. This might also feed into a "narrow" conceptualization of economic problems and solutions to them. As a remedy, Best (2007) calls for greater intellectual pluralism at the Fund and diversity in the recruitment and training of staff members. Recent changes in the Fund's lending arrangements, such as a more sympathetic approach to capital controls and greater space for Keynesian ideas, might signal a change in this regard (Chwieroth, 2014;Clift, 2018;Metinsoy, 2021). Nevertheless, this does not seem to have solved the IMF's legitimacy problems, either. This is not least because the change is fragmented, where ideational shifts in one policy area or Fund department may not spill over to another area to a high degree and existing and new practices co-exist for at least a period of time (Kaya & Reay, 2019). Some scholars have particularly drawn attention to the gap between the rhetoric of change and the continuation of existing practices, which they called "organized hypocrisy" at the Fund, referring to the deliberate discrepancy between rhetoric and practice (Kentikelenis et al., 2016). This gap undoubtedly harms the Fund's legitimacy in the eyes of borrowing governments and their citizens.
Finally, the Fund has been criticized in terms of the consequences of its programs, such as its negative impact on economic growth (Przeworski & Vreeland, 2000), exacerbating inequality (Garuda, 2000;Lang, 2021;Oberdabernig, 2013), lowering labor's income (Vreeland, 2002), diminishing labor rights (Caraway, 2006;Reinsberg et al., 2019), and triggering civil wars (Hartzell et al., 2010) and coups d'état (Casper, 2017). As a response, the Fund has been promoting the idea of "country ownership" (Best, 2007;Seabrooke, 2007) and greater attention to the "political economy" of the borrowing country (author's interview with a senior IMF official, 2021; see also Nunn & White, 2016). This, however, assumes that programs are designed with certain transparency and objectivity and without informal political influence behind closed doors, such as the influence of the US. This article aims to delve deeper into why and how conditionality that leads to adverse economic outcomes, such as rising income inequality despite the declared aim of reducing it, may come about.
The article aims particularly to offer a cross-cutting perspective on the Fund's potential legitimacy deficiency and proposes to unpack the term "borrowing country." While quota reform may grant a greater voice to developing countries, it does not empower labor groups within borrowing countries to the same extent. Governments negotiate the programs at the Fund and can mobilize their alliance with the US to secure favorable deals for their own narrow interests. Scholars have previously demonstrated that the US engages in "informal, backdoor politics" at the Fund (Stone, 2008, p. 595) and that it secures larger loans and more lenient conditionality for its allies (Copelovitch, 2010;Dreher & Jensen, 2007;Dreher et al., 2015;Stone, 2008). Following the recent advancement in IMF studies in terms of unpacking conditionality (Dang & Stone, 2021), we can also look at what types of conditionality (in addition to the total number of conditions) US allies receive. Furthermore, greater attention to a country's political economy or greater intellectual pluralism may not be extensively helpful unless transparency and objectivity are ensured in the design of conditionality. This article looks at how governments, as negotiators of Fund programs, can use their seats at the table to signal their ideological proximity to the US and strengthen their geostrategic alliance with it by pushing for labor market reform under IMF programs. This need to "prove" ideological proximity is particularly strong for left-wing governments, which may be regarded more suspiciously by the US.
The next section discusses in more detail how an alliance with the US for left-wing governments may result in a larger number of labor conditions in IMF lending programs.

Governments, the United States Alliance, and the International Monetary Fund as "Scapegoat"
Governments often borrow from the IMF when they face balance-of-payments problems and when they cannot find credit on favorable terms in international private markets (Copelovitch, 2010). They sometimes, however, go to the IMF when they also want to conduct costly reforms at home and lack the power base to execute them (Vreeland, 2006). In such cases, they use the IMF as an external anchor and as a "scapegoat" to bypass domestic opposition and shield themselves from criticism from domestic constituencies (Vreeland, 2006). They then negotiate with the Fund with the purpose of implementing those politically costly reforms at home. Finally, US allies are more likely to borrow from the Fund due to the privileged position of the US simply because their request for loans are more likely to be approved by the Fund (Stone, 2008).
Scholars have previously demonstrated that governments have some influence on the design of conditionality. For instance, powerful groups such as the military might avoid budgetary cuts under IMF programs thanks to their connection to and representation by the government, while other less organized groups such as the education sector or labor groups might suffer from extensive budgetary cuts (Caraway et al., 2012;Nooruddin & Simmons, 2006). Affirming this point, a senior official in the Greek government, who negotiated a program with the IMF in 2012, argues that they had some wiggle room to oppose some conditionality and protect some groups such as the education sector from cuts (author's interview with a senior Greek official, 2014).
We can logically assume that governments might want to hold the discretion of how to distribute the burden with minimum political cost and with the greatest benefits to themselves. Left-wing governments, however, face a dilemma because of split domestic and international costs and benefits. Domestically, they are often elected on a labor-friendly platform, and their mandate is often not to curtail labor rights but to expand them. Therefore, we can assume that labor market reforms are highly costly, especially for left-wing governments. Internationally, their pro-labor left-wing agenda may provoke suspicions in the international market and can risk their geopolitical alliance with the US. They can solve this dilemma by externalizing the blame and "scapegoating" the IMF for labor market reforms in domestic politics. If they can blame the Fund for those reforms, this exonerates them from the burden. Internationally, they can demonstrate to the US and international markets that they do not have a "radical" leftist agenda. To put it differently, they show that they are "good leftists" and ideologically close to the US. For instance, in Brazil in 1999, the left-leaning government received five labor conditions with an above-average alignment score with the US (0.26) and did not receive any labor conditions in 2002 when its alignment score went down (0.18). In Eastern European borrowers of the Fund-such as Romania in 2003, Macedonia in 1997, and Hungary in 1997-labor market reform under IMF programs may also signal a break from the Soviet past. This break with the past coincides with an alliance with the US and a greater number of labor conditions. Hungary is an interesting example with six labor conditions and an alignment score of 0.46 in 1997.
Although perhaps every government may want to hold the ultimate discretion on reforms and have the greatest benefits with the minimum political costs for themselves, they may not always have access to this type of deal. Due to a certain power asymmetry between the Fund and the borrower, borrowers do not have extensive bargaining power. The Fund is the creditor, and the borrower is naturally in need of the loans and hence in a weaker position. One can argue that the Fund can push for its agenda of reducing income inequality despite governments' different priorities and geostrategic calculations. US allies, however, may overcome this type of asymmetry with the Fund thanks to the US privileged position at the IMF. The US maintains an effective veto power since it holds more than 16% of the votes and 85% of the votes are required for qualified majority decisions (Woods, 2006). Furthermore, it commands an informal influence over the institution (Stone, 2008). Hence, the US might help its allied governments to influence the design of conditionality and help governments execute labor market reforms under an IMF program without much cost to themselves.
Indeed, there is a statistically significant positive correlation between voting in line with the US in the UN General Assembly (a proxy for the alliance with the US) and the relative frequency of labor conditions for IMF borrowers (0.61, p = 0.02). Data on UN General Assembly voting come from Dreher et al. (2015) and on IMF conditions from Kentikelenis et al. (2016). On the other hand, there is a statistically significant (although substantively small) negative correlation between an alliance with the US and the relative frequency of fiscal conditions in programs (−0.07, p = 0.03). Furthermore, delving deeper into the data, we see that US-allied left-wing governments receive more labor conditions and fewer fiscal conditions, while US-allied right-wing governments receive fewer conditions on both accounts. Figure 1 below shows how labor conditions increase for left-wing governments, as the alliance with the US becomes tighter (left panel). It also shows that labor conditions decline for US-allied right-wing governments (right panel).
Furthermore, without the US alliance, there is not a statistically meaningful relationship between labor conditionality and government partisanship. Left-wing governments with a looser alliance with the US receive on  average fewer and less stringent labor conditions. One can argue that, without the need to project ideological proximity to the US, these governments are more likely to protect labor interests at the negotiation table. They, however, receive a greater number of fiscal conditions and a greater number of total conditions, confirming earlier studies that the US intervenes to secure more lenient programs for its allied governments in terms of total number of conditions. In other words, a strategic alliance with the US alters the distribution of conditions and the choice between labor and fiscal conditions for left-wing governments. An alternative explanation might be that left-wing governments have a short-term electoral strategy: By keeping fiscal conditions to the minimum, governments preserve their ability to govern and distribute strategic resources to their constituencies in order to shield them from the impact of the crisis and shrinking resources. Secondly, via labor conditions, they socialize and disseminate the impact of conditions to all parts of the society and reduce the weight on their electoral constituencies. In other words, they trade labor conditions with fiscal conditions in service of their re-election strategies. However, if this is the case, we should observe a similar calculation for all left-wing governments. In the design of conditionality, we see that, without an alliance with the US, left-wing borrowers of the Fund receive fewer and less stringent labor conditions. In other words, labor conditions might really be the price of the alliance with the US.
Another alternative explanation may be that left-wing governments find it easier than their right-wing counterparts to build a coalition for reform due to general support for pro-market reforms from the right-wing opposition (Beazer & Woo, 2016). Right-wing governments, on the other hand, face fierce opposition from the left-wing parties when they want to initiate reform. This, however, contradicts the fact that non-US-allied left-wing governments receive fewer labor conditions. Furthermore, the theory may be attributing more agency to the opposition than the government itself in policy initiation and implementation. The empirical data seem to show that US-allied left-wing governments are more likely to receive labor market conditions than right-wing and centrist governments.
The IMF's labor conditionality can be highly intrusive, overhauling the labor market regulations. Some common labor conditions consist of dismantling collective labor agreements and replacing them with firmand individual-level agreements, such as in the Greek and Portuguese programs in 2010 and 2011, respectively (IMF, 2010a(IMF, , 2011. This naturally diminishes the bargaining power of labor groups and results in a reduction in their income. Conditions can also directly lower the minimum wage and pension rights as in Latvia in 2010 (IMF, 2010b). They can mandate the layoff of public sector workers (Rickard & Caraway, 2019), foster privatizations (Caraway et al., 2012), and lower the wages of pub-lic sector workers (IMF, 2010b). Via the "demonstration effect," this reduces wages in the private sector as well (IMF, 2009). Conditions can reduce the severance payment and unemployment benefits and make firing easier in general. The maximum duration of temporary contracts can be extended, and the number of hours one can work on a part-time contract might be prolonged (IMF, 2010a).
Labor conditions in IMF programs often contribute to rising income inequality. Especially dismantling collective bargaining rights results in an increase in the income of wage earners at the top, while those of lower-income groups diminish further (Wallerstein, 1999). With lowered security due to layoffs from the public sector and easier firing conditions in the private sector, workers accept lower wages. Scholars have found that IMF programs in general increase inequality (Forster et al., 2019;Oberdabernig, 2013). Furthermore, they found that this effect is mainly driven by income losses among lowerincome groups (Lang, 2021). Labor conditions lower the income of mainly vulnerable groups in society. This squarely contradicts the Fund's declared aim of reducing inequality and diminishes the Fund's legitimacy due to the discrepancy between the rhetoric and policy.
According to the IMF, labor market reforms in its programs are geared towards bringing greater flexibility and efficiency to the labor market (IMF, 2013). They practically, however, reduce protection in the labor market. Considering the negative impact of those measures on the prospective and current income and security of labor groups, they are especially politically costly for left-wing governments to implement. The IMF then becomes a convenient "scapegoat." Being "scapegoated" in this way has an additional negative effect on the legitimacy of the Fund in two ways. Governments "scapegoat" the IMF using "elite cues." Research shows that negative elite cues regarding international organizations stick more than positive messages, and this holds especially true for organizations such as the IMF, where public knowledge of the organization is limited (Dellmuth & Tallberg, 2021). "Scapegoating" diminishes the basis of the Fund's legitimacy in the public eye. This often translates into largescale protests and strikes, and the legitimacy of the institution is visibly diminished for future borrowers as well (Abouharb & Cingranelli, 2007;Pastor, 1987). Secondly, "backroom deals" harm the main basis of the IMF's legitimacy, i.e., its claim to non-political, expert knowledge (Best, 2007;Clift & Robles, 2021). Interference of politics such as the alliance with the US and the government partisanship in conditionality significantly damage the Fund's expertise-led legitimacy claims.
The IMF assigning a larger share of labor conditions for US-allied left-wing governments also harms procedural and output legitimacy, as discussed in the introduction of this thematic issue. Procedurally, the IMF is "instrumentalized" by governments, which lack the mandate and power base to implement politically costly reforms. The legitimacy of international organizations is sometimes argued to be granted by the willing participation of governments, who voluntarily sign up for these organizations (Guastaferro & Moschella, 2012). If those governments then use the international organization to circumvent the domestic opposition and shift the burden of adjustment on them in order to preserve their geopolitical alliances, then this naturally harms the procedural legitimacy claims. In terms of outcome legitimacy, rising inequality and diminished security and income for the majority of workers contradict the stated goals of the IMF, i.e., reducing inequality in its program countries. Additionally, it takes the blame for an outcome that a government is not normally elected to implement but can put in place (an extensive labor market reform) thanks to the anchor of the IMF.
The next section provides the empirical support for the theory proposed in this section that US-allied left-wing governments receive a larger number of labor conditions than non-US allies, controlling for preexisting labor market regulations and firing costs and other political and economic variables.

Quantitative Evidence: Alliance with the United States, Government Partisanship, and Labor Conditions
The proposed theory in this article predicts that left-wing governments, allied with the US, will receive more labor conditions in their programs. In order to test this theory, I draw on a sample of IMF borrowers between the years 1980 and 2014. The unit of analysis is country-year following earlier studies in the field (Caraway et al., 2012;Copelovitch, 2010;Dreher & Jensen, 2007;Nelson, 2014). I limit the sample to countries that have been under an IMF program for five or more months in a particular year in order to avoid false "zeros" on labor conditionality for non-borrowers (naturally, non-borrower countries do not receive any labor conditions). It can also be argued that some systematic commonalities among IMF borrowers may require them to implement labor conditions. For this reason, in addition to limiting the sample to IMF program countries, I control for selection into IMF programs: 1 indicates if a country has been under an IMF program for five or more months in a given year and 0 indicates otherwise. In the selection, I look at the impact of GDP (economic size), GDP per capita income (relative wealth), GDP growth (economic crisis), current account deficit as a ratio of GDP (balance-of-payment imbalances), and external debt as a ratio of gross national income (GNI). All variables come from the World Bank World Development Indicators data set. Finally, I also control for recidivism since repeat borrowers are more likely to go back to the IMF. I measure recidivism as the average number of programs in the past five years.
Two-stage models for count dependent variables, such as labor conditions in IMF programs that account for sample selection, are relatively new in the literature. For this purpose, I use an R package specifically developed for count data (Wyszynski & Marra, 2018). The model fits a probit selection model in the first stage (the dependent variable is "selection into the IMF") and a negative binomial model in the second stage (the dependent variable is the number of labor conditions). I use a classic Heckman selection model when the relative frequency of labor conditions is the dependent variable (since it is a continuous variable ranging from zero to 0.5).
Data on the main dependent variable-"labor conditions"-come from the Kentikelenis et al.'s (2016) data set on IMF conditions. I measure labor conditionality in three different ways for robustness checks: (a) count of labor conditions in a program, (b) weighted stringency of labor conditions with greater weight assigned to stricter conditions such as prior actions and performance criteria and less weight to structural and indicative benchmarks, and (c) the relative frequency of labor conditions within the total number of conditions (i.e., labor conditions divided by total number of conditions). The third measure is a safeguard in case larger programs also contain a higher number of labor conditions due to the greater adjustment requirement. For the second measure, the performance criteria and prior actions are given higher stringency since their fulfilment is a requirement for the release of the IMF's tranche. Non-fulfilment of a benchmark does not necessarily result in holding up the tranche. Descriptive statistics for all variables in the analysis are in the online Supplementary Material.
Following the earlier robust literature, I proxy the "alliance with the US" with a measure of the UN General Assembly voting (Copelovitch, 2010;Dreher, 2006;Dreher & Jensen, 2007;Dreher et al., 2015;Stone, 2008). US allies follow the US voting patterns in the UN General Assembly (Dreher, 2006). Higher values on the UN voting variable indicate greater alignment of voting between the country and the US and hence a closer alliance between the two. Data come from Dreher et al. (2015). Data on "left-wing governments" come from the Database of Political Institutions (Cruz et al., 2018) and is coded as 1 if the incumbent government is left leaning and 0 for right-wing and center governments. In the analysis, I look at the interaction between the two, that is left-wing incumbent government and the alliance with the US. The list of left-leaning governments, the number of labor conditions, and their alignment score with the US in the UN General Assembly are in the online Supplementary Material.
In the second stage of the analysis, I also control for several political and economic variables that might affect labor conditions. First, preexisting strict regulations in the labor market might prompt a greater number of labor conditions. Conversely, in an already flexible labor market, there may not be a great need for a labor market reform. In order to control for the impact of a "regulated labor market," I add a composite variable based on legal protection of employment and safeguards against overtime work. The measure includes indicators for firing costs, collective agreements, and wage protection, and the data come from the Centre for Business Research's Labor Regulation Index. The variable is lagged for one year in order to exclude the potential impact of the IMF on labor market regulation.
Secondly, in "election" years, governments may be more likely to shy away from labor conditions, and hence we may observe fewer conditions then. Data on elections come from Database of Political Institutions, and 1 indicates that there was an election in a particular year and 0 indicates otherwise. Relatedly, "democracies" may be more representative of and responsive to labor interests and might avoid labor conditions. Data on democracies come from the Polity II project. On a 20-point scale, 0 indicates an authoritarian regime and 20 indicates a democratic regime. Finally, there has been an upward trend in IMF's labor conditions starting from the 1980s until recently (Caraway et al., 2012). In order to capture this trend, I add a "time trend" variable to the analysis. Table 1 below reports the results for selection into IMF programs in the first stage and determinants of labor conditions (for two different measurements), accounting for selection into the IMF, in the second stage.
The selection into IMF programs seems to follow the pattern identified by earlier studies. Countries undergoing economic difficulties, such as negative GDP growth, and poorer countries with lower GDP per capita and rising external debt as a proportion of GNI are more likely to borrow from the Fund. Similarly, repeat borrowers go back to the Fund. Countries with larger economies, measured by GDP, are more likely to borrow from the Fund but the substantive impact is very small. A current account deficit, on the other hand, does not seem to have a statistically significant effect.
The results also show that left-wing governments, when the alignment with the US is zero, receive fewer labor conditions. When the left-wing government is zero (i.e., when the government is right-leaning or centrist), increasing US alignment results in a greater number of labor conditions. Furthermore, as the interaction term shows, an increasing alignment with the US for left-wing governments increases labor conditions both as a count (Model 1) and in terms of strictness of labor conditions (Model 2). Conversely, right-wing governments and non-US allies comparatively are less likely to receive labor conditions. A two-unit increase in the US alliance variable results in almost three and a half extra labor conditions for left-wing governments. Furthermore, conditionality becomes stricter for US-allied left-wing governments, such as receiving more performance criteria and prior actions in their programs. The interaction term is statistically very significant for both models.
In addition, as stipulated above, elections reduce labor conditions, all else being equal. Governments probably fear the negative consequences of labor conditions in election years. Interestingly, democracies receive more labor conditions. This might be because the measure captures some of the effects of the alliance with the US as democratic governments have been more likely to be the US allies since the Cold War period. The time trend captures the increasing number of labor conditions in IMF programs throughout the years. Finally, regulated labor markets receive fewer labor conditions, while, paradoxically, more flexible labor markets are more likely to receive conditions. In regulated labor markets, labor groups may be powerful and block flexibility measures and IMF labor conditions. This seems to provide additional support for the theory that conditionality is not purely determined based on objective economic conditions. Results for the Heckman selection model are in the online Supplementary Material.

Robustness Checks
In order to check the robustness of results, I fit several different regression models. I particularly look at the results with frequently used modelling of negative binomial regression with robust standard errors clustered across countries (Caraway et al., 2012; Model 1 in Table 2), negative binomial regression for panel data with fixed effects (Model 2 in Table 2), and ordinary least  Table 2). Fixed effects in the models control for heterogeneity of countries in the sample. Statistically significant results in these models yield strong support for the proposed theory. Table 2 reports the results. All three models in Table 2 show that the interaction term between a left-wing government and the alliance with the US remains significant. When the incumbent government is a left-wing one, one unit increase in the voting alignment with the US increases the likelihood of receiving labor conditions approximately by 48%, ceteris paribus. When the alliance with the US is zero, i.e., when there is a complete divergence between the US and the given country, then having a left-wing incumbent government would reduce the likelihood of receiving labor conditions by 31%. This confirms the theoretical assertion that governments allied with the US can receive "backroom deals." Furthermore, all models show that left-wing governments receive fewer labor conditions and a smaller share of labor conditions in the total number of conditions when the alignment with the US is zero.
In addition to those robustness checks, I also run a placebo test investigating if the impact of the independent variable is due to a "placebo effect" and the model merely predicts the number of conditions in programs. Table 3 reports the results, where the total number of conditions is the dependent variable. Table 3 shows that the impact of the interaction term between left-wing government and alliance with the US on labor conditions is not due to placebo effects. They do not influence the total number of conditions in the pro-grams. Neither of the left-wing or US alliance variables reaches statistical significance.

Conclusion
This article has discussed the IMF's efforts to garner greater legitimacy by incorporating issue areas such as reducing inequality in its lending programs. Scholars have previously argued that the bureaucratic culture of the Fund, its narrow economic expertise, and lack of political-economic understanding of local conditions in borrowing countries explain its limited success in this regard. This article aimed to bring an additional perspective by looking at the role of the alliance with the US. The US occupies a privileged position at the Fund and can affect lending decisions as well as the design of programs. It can secure favorable terms for its allies at the Fund. The article showed the evidence that indeed the US provides more lenient conditions for its allied governments while helping to shift the burden of adjustment onto labor groups. I argued that the impact is especially prevalent for left-wing governments, which would face high political costs if they launched a labor market reform on their own. Left-wing incumbent governments signal their ideological proximity to the US by launching labor market reform under an IMF program, strengthening their geostrategic alliance with the greatest stakeholder at the Fund while "scapegoating" the IMF in the process in order to avoid political costs associated with labor market reform. The article contributes to the earlier studies that demonstrated that the US allies receive more lenient terms from the Fund. It argued that Notes: Negative binomial regression for panel data with fixed effects; robust standard errors in parentheses; *** p < 0.01, ** p < 0.05, * p < 0.1. while governments as institutions might benefit from the "selective friendship" of the US, labor groups may not. These findings have significant implications for the IMF's legitimacy. Procedurally, the Fund assigning more conditions to allied left-wing governments not only allows governments and international powers such as the US to instrumentalize the IMF, but it also provides a "false target" for domestic audiences, i.e., the IMF rather than the government. In terms of outcomes, it exacerbates inequality in domestic politics and potentially reduces the success of IMF's programs. Greater transparency in the design of programs and autonomy for the IMF as an institution from its principals, and especially from the US, might help improve the institution's legitimacy. Scholars can study if the Fund takes "real" steps towards reducing economic inequality in borrowing countries in the future, drawing on the findings of this article. About the Author Saliha Metinsoy is an assistant professor of international political economy at the University of Groningen, in the Netherlands. She works for international organizations with a particular focus on IMF programs, labor markets, and international finance. Her work has previously appeared in the Journal of Public Policy and Global Perspectives. Dr Metinsoy holds a PhD in politics and international relations from the University of Oxford.