Governments as borrowers and regulators

Abstract

The ability to borrow is important for government survival. Governments routinely resort to policies that privilege their own debt on financial markets, exploiting their dual role as borrowers and regulators. We label such policies as borrowing privileges. These borrowing privileges nudge investors to hold the government’s own debt. They share similarities with prudential regulation, but skew the market in favor of the government’s debt; and they share similarities with financial repression, but are less severe and thus consistent with the growth of financial markets. Introducing the first systematic dataset documenting the use of such policies across countries and over time, we demonstrate that governments implement borrowing privileges when their interactions with the global economy heighten fiscal needs: when borrowing costs indicate tightened access to credit, when trade liberalization undercuts revenue, and where fixed exchange rates increase the value of fiscal space. Despite the mobility of financial assets and constraints from global markets, governments retain latitude in regulating domestic markets to their own fiscal benefit.

How governments finance their operations is at the core of debates on the distribution of power within and across societies, the development of state institutions, and the political control of the economy. To raise funds, governments routinely borrow from banks and financial markets. A central concern of governments is thus to ensure their debt is in demand (Schultz & Weingast, 2003; Stasavage, 2011; Poast, 2015; DiGiuseppe & Shea, 2016; Bunte, 2019). If investors are unwilling to hold government debt, financing dries up and borrowing costs increase, limiting the fiscal space of governments and threatening their survival. How, then, do governments generate and sustain demand for their debt?

Existing work emphasizes how the need to attract credit places limits on government autonomy, both through self-imposed institutional constraints (North & Weingast, 1989; Broz, 1998; Dincecco, 2011) and through structural constraints imposed by capital mobility (Mosley, 2000; Clark, 2003; Plümper et al., 2009; Bodea & Hicks, 2015; Betz & Kerner, 2016; Pond, 2018; Ballard-Rosa et al., 2021; Zeitz, 2022). That investors can shift to other investment opportunities constrains governments, drives convergence toward market-friendly policies and institutions, and ensures continued repayment.

Pointing to the dual role of governments as borrowers and regulators, we argue that governments retain substantial policy autonomy over domestic financial markets, and that they use this autonomy to put their own debt in a privileged position on domestic markets during times of heightened fiscal needs. While limits to government autonomy can help governments attract credit, governments have a more direct regulatory mechanism available: They can encourage or mandate that financial actors hold their own debt, resorting to policies we label borrowing privileges in the following.

A wide variety of policies fall under this label. Examples are policies that require banks or institutional investors, such as pension funds, to direct a certain percentage of their assets toward the government’s own debt; policies that treat the government’s own debt more favorably than other assets; and policies that require an authorization to purchase assets other than the government’s own debt. Notably, these policies create a privileged position for the government’s own debt relative to other assets, including relative to debt issued by other governments.

Such policies have long been noted in the literature. For example, both Calomiris and Haber (2014) and Reinhart and Sbrancia (2015) identify several of these policies and consider them as instances of fiscally motivated financial repression. At the same time, several of these policies have been understood as forms of prudential regulation elsewhere (e.g., Alam et al. 2019). This contrast underscores that these policies remain poorly understood and are rarely considered as an explicit policy category, in part because there has been no attempt to provide a systematic account of the breadth and existence of borrowing privileges.

In this paper, we therefore make two main contributions. First, we identify borrowing privileges as a distinct class of policies implemented by governments for fiscal motivations. Drawing on an original data set of 87 countries from 1996 through 2015, we present a systematic account of borrowing privileges. We document that many governments – especially governments in middle-income countries, where sufficiently developed financial markets coincide with fiscal needs – implement policies that privilege their own debt over other assets through a wide variety of measures. These policies shore up demand for government debt by banks and institutional investors. Borrowing privileges take an intermediate place between prudential regulation and financial repression. They share similarities with prudential regulation, but skew the market in favor of the government’s debt; and they share similarities with financial repression, but are consistent with the growth of financial markets.

Borrowing privileges offer several upsides to governments. They guarantee demand for government debt from creditors who might otherwise look for investment opportunities elsewhere. They allow governments to balance their fiscal needs with the benefits of growing financial markets. And they are politically expedient: Both the immediate cost and the distributional consequences are opaque from the perspective of most citizens – who suffer through lower investment returns, suppressed credit markets, and limited portfolio diversification. In related work, we further demonstrate how the adoption of borrowing privileges is driven by domestic politics: Governments reach for borrowing privileges especially under democratic institutions, where such intransparent policies are particularly attractive to satisfy heightened fiscal needs (Betz & Pond, 2023a).

Second, we demonstrate that governments turn to domestic markets as a source of credit when their interactions with global markets indicate tighter credit conditions and put governments under fiscal pressure: when sovereign bond yields increase; when trade liberalization reduces government revenue; and when fixed exchange rates limit monetary autonomy and make fiscal autonomy more important. The combination of these measures increases our confidence that we capture fiscal need and not, for example, some underlying model of economic governance. While we cannot provide a research design to arrive at causal effects, these results underscore how governments rely on intransparent and distortionary policies when their access to credit is tightened and fiscal space becomes more important.

Our results highlight that governments retain substantial latitude in regulating domestic markets to their own (fiscal) benefit, compensating for constraints imposed by international markets. Financial assets may be mobile, but financial services, which are tied to immobile customers in the local market, are not. This induced immobility creates room for governments to regulate financial service providers, including banks and institutional investors. By imposing regulations that require or encourage investors to hold the government’s own debt, governments can inflate demand for their debt – even in the short term, which contrasts with much of the existing literature in political science that focuses on long-term institutional solutions and structural constraints on government autonomy.

Beyond contributing to the literature on economic globalization and government autonomy, our paper speaks to a large literature on financial regulation. We add to this literature in several ways. Empirically, existing studies of financial regulation frequently resort to indirect measures derived from economic outcomes, such as the presence of negative real interest rates, which conflate regulatory policy with market-driven incentives and behavior. We provide a systematic measure of government policies, rather than economic outcomes, across a large set of countries. This allows us to distinguish the incentives of banks to hold government debt from mandates by governments that require banks to do so.

Theoretically, some work on financial regulation emphasizes distributional consequences: governments use financial regulation to please market incumbents, in exchange for lobbying contributions, or to channel credit to politically connected firms – and in the process they often undermine financial development (Perotti & Volpin, 2004; Pinto et al., 2010; Pepinsky, 2013; Menaldo, 2016; Naczyk & Hassel, 2019). Other accounts emphasize the trade-off between financial stability and competitiveness in financial regulation (Singer, 2004; Copelovitch & Myren, 2018; Copelovitch & Singer, 2020).

Instead of distributive politics or the trade-off between stability and competitiveness, we join a literature that emphasizes the government’s fiscal needs as a driver of regulation (Bastiaens & Rudra, 2016; Reinhart & Sbrancia, 2015; Menaldo, 2016). Complementing this literature, we highlight how the government’s fiscal needs, and as a consequence the regulation of financial markets, are driven by a government’s interactions with international markets; we document how these measures are compatible with functioning financial markets, rather than stunting their development through blatant repression; and we show how governments tap a large pool of resources provided by institutional investors, including pension and insurance funds. Moreover, we contribute to a growing literature which emphasizes the political implications of several aspects of the ownership of government debt, including its currency denomination, its issuance on domestic versus international markets, and the identity of creditors (e.g., Bunte 2019). We contribute to this literature by focusing on and measuring a specific category of financial regulation through which governments shape who holds government debt.

Finally, household borrowing, stock market investment, and the privatization of welfare states have produced rapid growth and widespread participation in financial markets (Brooks, 2007; Kerner, 2020), which suggests pressure on the policy autonomy of governments (Freeman & Quinn, 2012; Pond & Zafeiridou, 2020). Yet, where financial assets are employed in the provision of financial services, they are specific to local markets. This creates regulatory space. Market incumbents are better off staying in the market and accepting modest amounts of regulation than abandoning established markets. We thus take a more optimistic view of the ability of governments to regulate markets, joining several recent studies that extend how we think about capital mobility (Johns & Wellhausen, 2021; Bauerle Danzman & Slaski, 2022; Chen & Hollenbach, 2022; Betz & Pond, 2023b). Borrowing privileges are attractive to governments not despite the growing importance of financial markets, but because this growth has created a market to tap.

1 Government privilege

Governments can employ a relatively simple tool for securing demand for government bonds: They require or incentivize banks and institutional investors – like insurance companies or pension funds – to hold a share of their assets in the government’s own debt, imposing policies that we call borrowing privileges. Relative to other assets, including the debt of other governments, these policies create a privileged position for the government’s own debt on financial markets.¹

As a specific example of such policies, in 1997 the government of Uruguay mandated that “Private pensions must hold a minimum ratio of total assets (from 30% to 60%) in the form of government securities” (1998 IMF AREAER Report). For comparison, common recommendations for retirement planning suggest that someone who is 40 years old should invest between 20% and 40% of their retirement assets in diversified bond portfolios, leaving investment shares significantly below 30% for bonds of their own government. Similarly, banks on average hold about 9% of their assets in government bonds, only a portion of which are bonds of their own government (Gennaioli et al., 2014). Restrictions such as in this example from Uruguay thus imply a significant increase in government bond holdings compared to what banks and institutional investors would hold in the absence of regulation.

Borrowing privileges are not always outright mandates to hold government debt. For example, South Africa implemented another type of borrowing privilege. To stabilize banks, governments often maintain liquid asset requirements. Ideally, these ensure that banks and the government are not caught with insufficient liquid assets to meet liabilities. In South Africa from 2010 to 2015, “government securities, treasury bills, land bank bills, and reserve bank securities” qualified as liquid assets (2011-2016 IMF AREAER Reports). This policy privileges South Africa’s debt, as holding sovereign debt is more lucrative than holding cash: government bonds pay interest, whereas cash pays no interest and loses value due to inflation.

As another example of a less immediate privilege for government debt, in 2006, the Mexican government required that all purchases of foreign securities be authorized by the Mexican government. This authorization requirement applied to banks, securities firms, insurance companies, and privately managed pension funds. Exempt from these requirements were “securities issued in foreign currency by the federal government” (2007 IMF AREAER Report). By exempting government securities, the government ensured that it was easier to purchase government debt than other stocks and bonds.

We argue that governments impose these policies for their own fiscal benefit, a common but frequently overlooked motivation for financial regulation (Calomiris & Haber, 2014; Reinhart & Sbrancia, 2015). As Reinhart and Sbrancia (2015, 291) put it, “It apparently has been collectively forgotten that [financial regulation] played an instrumental role in reducing [...] massive stocks of debt” since World War II. By requiring banks and institutional investors to hold government debt, borrowing privileges ensure a steady investor base and drive down borrowing costs. For example, an IMF study found that increasing the share of government debt held by institutional investors by 10 percentage points reduces yields on government bonds by about 25 basis points (Andritzky, 2012). To put these effects into perspective, at 2018 debt levels, a reduction in interest rates of this magnitude corresponds to savings in annual interest payments of over US$800 million for Argentina, of US$1.4 billion for Mexico, and of US$3.2 billion for Brazil. For governments with substantial borrowing needs, encouraging institutional investors to purchase government bonds promises to reduce their debt burden considerably.

These policies can be thought of as soft or hidden repression,² as they differ from more blatant forms of financial repression emphasized in the extant literature. First, financial repression that benefits governments fiscally is often understood as the implementation of interest rate ceilings, negative real interest rates (Giovannini & de Melo, 1993), or forced lending. Forced lending is often targeted toward politically connected firms or state-owned enterprises, which the government then might tax (Menaldo, 2016). Borrowing privileges, in contrast, benefit governments immediately. Second, borrowing privileges include policies that are more of a nudge than a requirement – as in the examples from South Africa and Mexico above. While the recent literature has noted that financial repression can include “directed credit to the government [through] regulatory measures requiring that institutions (almost exclusively domestic ones) hold government debts” (Reinhart & Sbrancia, 2015, 293), borrowing privileges can also be more subtle, in that they may stop short of a requirement to hold government bonds. Third, borrowing privileges apply to a broad pool of assets. In addition to the assets of banks, borrowing privileges can apply to pension and insurance funds. Fourth and in contrast to more severe forms of financial repression, borrowing privileges are compatible with the growth of financial markets.

For governments, borrowing privileges present trade-offs on several dimensions. First, incentivizing institutional investors to hold government bonds can drive down investment returns, which are passed on to citizens through lower yields on savings accounts and less diversified mutual funds or insurance products. At the same time, these are predominantly opportunity costs from the perspective of citizens. Compared to tax hikes, foregone investment returns are less likely to turn into political issues. The indirect costs of many financial products are relatively intransparent – analogous to high cash holdings in mutual funds or automated investment programs, which allow asset managers to earn profits while offering seemingly low-cost asset management. Indeed governments affect the visibility of the policies, and they have little incentive to implement disclosure requirements where unusually high allocations favor government debt. Even professional rating agencies have difficulty evaluating the implications of domestic financial regulation. In a 2017 report, Standard and Poor’s pointed out that it is difficult to forecast sovereign default rates “when domestic financial institutions hold the [government’s] debt" (S &P Global Ratings, 2017, 17).

These challenges in recognizing the costs of borrowing privileges are compounded by collective action problems. For each individual citizen, the costs of borrowing privileges are small. Citizens also frequently lack basic knowledge and pay little attention to financial regulation (Sokhey, 2017), although they may own financial securities through pension assets and investment accounts. And while banks and institutional investors are affected by these regulations, they might be able to pass at least some of the costs on to citizens. The political costs associated with taxation, and the diffuse and opaque costs of financial regulation, plausibly cause some policymakers to favor borrowing privileges over taxation.

Second, driving up holdings of government debt among domestic creditors can be an effective hands-tying strategy. While many citizens may not realize that they are creditors of the government during ordinary times, they certainly would notice their financial losses if the government defaulted on its debt. By raising the political cost of default, governments can increase the credibility of debt repayment, in turn allowing them to access financial markets more easily in the future. Thus, domestic creditors may constrain governments directly by intervening before governments engage in policies that steer them toward default (Stasavage, 2011); but they may also constrain governments indirectly, if default reveals information to those creditors. As described in Stasavage (2011), then, a stable base of domestic creditors increases confidence in the government’s willingness to repay its debt in the future, and this reciprocal effect in turn facilitates the government’s access to financial markets.

Yet, requiring large holdings of government debt is not without risk. Large holdings of government debt can present a major source of risk by “crowding out small borrowers, transferring risks to banks [...], and imposing risks on holders of pensions, annuities, and life insurance policies” (Hanson, 2007, 1). Borrowing privileges exacerbate these concerns by requiring financial institutions to hold more government debt than they otherwise would. Requiring large holdings of government debt may prevent creditors from balancing their portfolios in ways that mitigate risk, and they may strengthen the connection between government default risk and domestic financial instability. For example, exposure to government debt deepens the consequences of economic downturns (Gennaioli et al., 2014; Baskaya & Kalemli-Ozcan, 2016). Borrowing privilges can also have a destabilizing effect on financial systems. In 2013, Jens Weidmann, former President of the Deutsche Bundesbank and Chair of the Bank for International Settlements, noted that financial institutions should be restricted in holding their own government’s debt – while “governments are afraid of rising funding costs as a result of ending the regulatory privileges afforded to sovereigns [...] I do not think that this argument should keep us from doing the right thing” (Financial Times, September 30 2013).

The literature suggests two alternative explanations of financial regulation, based on distributive politics and on prudential motivations. First, borrowing privileges can create entry barriers into the financial market, providing rents to market incumbents. Second, borrowing privileges share similarities with prudential regulation. Governments frequently require that banks and institutional investors hold some portion of their assets in highly rated bonds, that they diversify their portfolios, and that the exchange rate denomination and maturity of their liabilities mirrors that of their assets. Where highly-rated corporate bonds are not available or rating agencies are unreliable, governments may rely on policies that require holdings of government bonds exclusively.

The U.S. Banking Act of 1863 and 1864 illustrates how both motivations can guide government behavior, alongside the fiscal motives emphasized here. The Act required national banks to have a minimum capital stock of $50,000 in towns with a population under 6,000, and of $100,000 in towns of up to 50,000; at least one third of the capital stock, or $30,000, had to be held in U.S. government bonds (National Monetary Commission, 1910). These requirements only applied to national banks, not to local banks. In smaller towns, it was almost impossible to earn a return on deposits that exceeded what could be earned on $50,000 invested elsewhere. Consequently national banks stayed out of local markets, allowing local banks to earn excess returns (Sylla, 1969; James, 1976). Thus, local banks benefited from borrowing privileges as an entry barrier.

The Act also had prudential components. National banks could only accumulate liabilities in proportion to their holdings of U.S. government bonds (at a rate between 90% and 100%), deposited at the Comptroller of the Currency. This can be interpreted as prudential regulation in a context where other safe assets were not available (and where even holding cash was not considered risk-free, given that currency issued by national banks was still a new form of paper money): a bank’s holdings of government bonds could be sold if the bank was no longer able to meet its obligations and the demand for cash redemptions.³

Importantly, however, the Act was also motivated by fiscal concerns: It required substantial purchases of government bonds by newly formed banks, expanding the demand for government bonds. In 1861, Treasury Secretary Salmon Chase described the advantages of creating “a large demand for government securities, of increased facilities for obtaining the loans required by the [civil] war” (cited in Robertson 1968, 38-39). In 1862, President Lincoln supported the policy in his annual message, highlighting “the steady market demand for Government bonds which the adoption of the proposed system would create” (cited in Robertson 1968, 42). Thus, as Sylla (1969), 659) notes, “Congress enacted the legislation primarily to increase the government’s borrowing power during the [civil] war by requiring all national banks to invest a portion of their capital in government bonds.”

The system created by the Banking Act remained in place until the creation of the Federal Reserve System in 1913.⁴ In the following, we explore the incentives of governments, including and beyond the U.S., to implement borrowing privileges during times of heightened fiscal needs, especially where their interactions with international markets drive those fiscal needs.⁵ We also acknowledge that government incentives plausibly complement each other: In augmenting their fiscal position, governments will likely prefer regulations that simultaneously reward their supporters and help stabilize the marketplace. In the empirical section and the online appendix, we strive to disentangle these motivations.⁶

1.1 Borrowing privileges in global markets

If borrowing privileges are at least in part adopted for fiscal reasons, we should observe the adoption of such policies when a government’s fiscal needs increase. In particular, we argue, as constraints from global markets indicate tightened credit conditions and heigthen fiscal needs of governments, governments turn to domestic markets as a source of credit.

Domestic financial markets have grown steadily over the last decades. Among the drivers of this growth have been the liberalization of financial markets, the opening of current accounts, and the widespread privatization of social security systems. Firms and households rely on financial markets to borrow and to invest (Wolff, 2017), and households increasingly turn to financial markets through privatized social security systems (Brooks, 2007; Kerner, 2020). With the increasing reliance of economies on financial markets, financial services now make up a significant share of economic activity (Greenwood & Scharfstein, 2013). What sets financial markets apart from other economic activity is that they provide an indirect revenue source: asset owners are potential creditors to the government. Thus, as domestic financial markets grow, they become attractive targets for government policy.

While financial assets pose regulatory challenges due to their mobility, financial services are often specific to local markets. Banking services, for example, are tied to domestic clients. Acquiring new clients presents a major challenge for banks. Exiting a market once these relationships have been established implies substantial costs, because the bank now has to compete with established market incumbents. A similar logic applies to other financial services, such as management of pension or insurance funds. This specificity to local markets creates room for market regulation. Market incumbents are often better off staying in the market and accepting modest amounts of regulation than relocating abroad and abandoning established markets. Consistent with this view, Standard & Poor’s observed, “a sovereign has flexibility [...] based on the unique powers it enjoys within its borders, particularly in its control of domestic financial and monetary systems” (S &P Global Ratings, 2017, 4).

Growth and regulatory space make domestic financial markets an attractive source of credit, especially as governments face increasing fiscal pressure from their interactions with international markets. We emphasize three sources of such pressure.

First, we expect governments to create borrowing privileges, seeking to increase demand for government bonds, when their costs of borrowing increase, indicating a slump in the demand for government bonds and reflecting tightened access to credit. Thus, increasing yields on government bonds, as indicators of declining demand and rising government borrowing costs, should be associated with the implementation of borrowing privileges. Because a substantial share of government debt is traded on global markets, the bond yield on international markets provides a clear indicator of the presence of fiscal pressure.

Of course, higher borrowing costs are not necessarily the root cause of tightened access to credit. They are an indicator of underlying problems, and of how markets perceive those problems and price them in. Moreover, while borrowing costs reflect fiscal pressure through tighter access to credit, they might not be driven by current fiscal problems. For example, borrowing costs might increase because of geo-political tensions or changes in a country’s economic outlook, but also due to changes in the international economic environment, such as inflation differentials between countries. We remain agnostic about the sources of changes in borrowing costs and simply maintain that higher borrowing costs, as a reflection of tightened credit access, lead governments to resort to borrowing privileges as intransparent policy instruments.

Second, borrowing privileges should become more attractive to governments that come under fiscal pressure from declining revenue sources. In the past decades, these were often the consequences of liberalization efforts that followed economic orthodoxy – including trade liberalization. Historically, trade taxes have been an important revenue source for many countries. While a small share of total tax revenue, even developed countries today still generate substantial revenue from trade taxes – U.S. tariff revenue in 2015 was approximately $40 billion, which is similar to the revenue from the federal capital gains tax on corporations, about 50% larger than the revenue from estate taxes, and four times the budget of the Department of the Interior. In 2020, a proposal to eliminate most import tariffs in Switzerland failed in parliament, in part because it was difficult to compensate for losing the revenue from import tariffs of about 500 million Swiss Francs.⁷

Demands for trade liberalization by both domestic constituents and membership in international organizations have eroded trade taxes as a main revenue source in many countries. At the same time, trade liberalization has led to demands for expanded government expenditure from those exposed to foreign competition, further increasing the fiscal strain on governments (Bastiaens & Rudra, 2018). Moreover, the lost revenue from tariff reductions has been difficult to replace with other taxes. Income taxes are politically unpopular and difficult to increase, while corporate taxes come under pressure from tax competition and profit shifting (Arel-Bundock, 2017). This heightens the attractiveness of borrowing privileges to governments. We thus expect that governments implement borrowing privileges as tariffs decline.

This is not to say that trade liberalization has only negative repercussions for government revenue. Trade liberalization, by increasing competition and the availability of inputs, can spur economic growth and thus broaden the tax base at least in some contexts (for a discussion, see Rodriguez and Rodrik 2000). Governments also might supplement protectionist policies with subsidies, imposing indirect fiscal strains. Trade liberalization might allow them to drop such subsidies. Nonetheless, trade taxes are difficult to replace with other revenue sources, and especially middle- and lower-income countries may face challenges in collecting other types of revenue. Recent evidence also suggests that trade liberalization tends to cause a loss in tax revenue (Arezki et al., 2021), and that many governments in the past decades were unable to compensate for the loss in trade taxes through other revenue sources (Cagé & Gadenne, 2018).

Third, fixed exchange rates place fiscal pressure on governments. Most governments have few capital controls in place. Those they have retained are largely ineffective. Capital mobility is now frequently considered a given (Frieden, 1991). Under open markets and fixed exchange rates, governments give up monetary policy autonomy. Without monetary policy at their disposal, they must rely on fiscal expansions as their main policy lever (Clark, 2003). Thus, governments have strong domestic political incentives to expand their fiscal space under fixed exchange rates: they need revenue to reward political supporters.

Additionally, with fixed exchange rates adjustment to international markets needs to happen internally, through changes in prices and wages (Manger & Sattler, 2020). Thus, fiscal policy becomes more important to smooth economic fluctuations in the aggregate, and to manage the political consequences of such adjustments. These demands for fiscal policy under fixed exchange rates should encourage governments to implement borrowing privileges.⁸

We summarize our expectations as follows.

Hypothesis 1 Governments should be more likely to implement borrowing privileges as they come under increasing fiscal pressure, and in particular

• when bond yields on government debt increase,

• when they liberalize trade policy, and

• when monetary policy constraints through fixed exchange rates increase the political value of fiscal autonomy.

This set of hypotheses combines conditions that are not easily reduced to a common dimension of economic governance. If borrowing privileges are part of a broader move to follow the economic orthodoxy, they should be associated with lower bond yields and lower tariff rates. Trade liberalization in particular is commonly an indicator of orthodox economic policies and the absence of rent-seeking governments. It follows standard policy prescriptions, including from international organizations such as the International Monetary Fund, for encouraging economic growth and market stability. Alternatively, if borrowing privileges are instead a form of financial repression and part of a departure from the economic orthodoxy, they should be associated with higher bond yields and higher tariff rates. We instead expect that fiscal pressure motivates policymakers to impose borrowing privileges. We anticipate that borrowing privileges are more attractive when policymakers come under fiscal pressure, as bond yields increase and as tariff rates decrease. The findings thus cannot be wholly explained by either following or departing from economic orthodoxy.

Before turning to the empirical evidence, we note that borrowing privileges are not merely a reflection of prevailing ideas of policies considered ‘best practices’ for regulatory policy. These generally encourage a diversification of asset holdings, not a concentration of holdings in a single government’s debt. The Basel Accords, which represent the global consensus on prudential banking regulation, propose risk-weighted capital adequacy standards. These standards identify reserves that banks must hold to alleviate risk, but they do not mandate that these reserves be held in government securities. The Basel Accords provide preferential treatment for holdings of sovereign debt – by considering highly-rated government bonds as risk-free (or, in the case of Basel I, OECD debt more generally) – but this preferential treatment is not limited to debt of the domestic government.⁹

2 Empirical evidence

We create a systematic measure of borrowing privileges for 87 countries annually between 1996 and 2015.¹⁰ The measure captures regulations that favor the government’s own debt over other assets, and that therefore encourage private domestic financial actors to hold the government’s debt. We coded the measure using information in the IMF AREAER Reports, which are a common source of information on economic policy, including for most measures of capital account openness and the exchange rate regime (see Quinn and Inclan 1997; Chinn and Ito 2008; Karcher and Steinberg 2013).¹¹ Two coders independently read the annual reports and coded the presence of policies that privilege government debt. When discrepancies arose, we revisited the reports to reconcile the coding.

Policies are coded as borrowing privileges only if they privilege the government’s own debt over other assets, including over the debt of other governments. Stipulations to hold assets rated highly by rating agencies, government debt of OECD members, or only domestic or domestically denominated debt are not considered borrowing privileges. Regulations that do not explicitly favor the government’s own debt are therefore not included in our measure of borrowing privileges.

The countries included in the sample are shown in grey in Fig. 1 and listed in the appendix. Countries in light grey never implemented borrowing privileges. Countries in dark grey implemented borrowing privileges for at least parts of the sample period. The sample represents countries from all regions of the world and includes all countries that were, by the end of the sample period, considered emerging markets by J.P. Morgan’s EMBI Global Index, the leading benchmark for emerging market government bonds, or by the IMF (with the exception of Taiwan, for which no IMF AREAER reports are available). The sample also includes all members of the OECD.

Fig. 1

Countries included in the sample. Light grey: never implemented borrowing privileges. Dark grey: borrowing privileges for at least some of the sample period

Fig. 2

Share of countries with borrowing privileges (dark grey, solid line) and with capital account restrictions (light grey, dashed line). 87 countries, 1996-2015

2.1 Descriptive overview

Before turning to the empirical analyses, we provide a descriptive overview. In addition to describing the use of borrowing privileges across countries and over time, we also situate them in the broader policy context emphasized in the literature on financial regulation, between prudential regulation and economic orthodoxy, on the one hand, and financial repression, on the other hand.

Over a quarter of the countries in our sample – 25 out of 87 countries – implemented some type of borrowing privilege during the sample period. Borrowing privileges occur predominantly in non-OECD countries. Only three OECD countries, Portugal, Mexico, and Turkey, had borrowing privileges in place for at least part of the sample period. To be sure, in OECD countries, stipulations to hold highly-rated assets or debt from OECD countries are common (and part of standard prudential regulatory frameworks). But these stipulations do not fall under our more conservative definition of borrowing privileges: investors retain discretion in complying with these regulations, they do not privilege the government’s own debt over other assets, and they are plausibly motivated by prudential concerns, and indeed part of regulatory frameworks like the Basel Accords.

The incidence of borrowing privileges increased over time, as shown by the solid line in Fig. 2. Despite the overall trend toward more borrowing privileges, the implementation of these policies is frequently reversed. While borrowing privileges were newly implemented 27 times during the sample period, they were also removed 18 times.

In the aggregate, the implementation of borrowing privileges coincides with a trend towards liberal economic policies. The light grey, dashed line in Fig. 2 displays the proportion of countries in the sample with closed capital accounts (CAP100 scores of 50 or lower, updated data from Quinn 1997). While governments have liberalized capital accounts over the past two decades, an increasing share of governments created borrowing privileges for their own debt. Among non-OECD countries, at the end of the sample period, almost one in three countries had borrowing privileges in place.

We next compare the presence of borrowing privileges at different levels of development. Figure 3 displays the density of countries in the sample by log GDP per capita (data from the World Bank), log stock market capitalization to GDP, and log private banking credit to GDP (data from the Global Financial Development Database).

Fig. 3

Kernel density estimates of log GDP per capita (left panel), log stock market capitalization to GDP (middle panel), and log private bank credit to GDP (right panel) for countries with (solid) and without (dashed) borrowing privileges

Borrowing privileges seem to follow a non-linear pattern; they are more common in middle-income countries, than in low- or high-income countries, and they are more common at middling levels of financial market development and bank savings. Low-income (capitalization and savings) countries have smaller markets to tap, making borrowing privileges less useful. High-income countries plausibly have sufficient demand for government debt, making borrowing privileges less necessary. It is in middle countries where borrowing privileges are the most attractive: markets are sufficiently large to provide meaningful revenue to governments; at the same time, they are sufficiently small that market incentives alone are inadequate to satisfy government demands for revenue.

To put borrowing privileges in the context of other types of financial regulation, we compare borrowing privileges to financial repression and to prudential regulation. Figure 4 shows that borrowing privileges were predominantly implemented in countries that in the past had closed capital accounts and dual exchange rates (exchange rate data from Ilzetzki et al., 2019), at a time when both policies ran counter to the economic policy consensus and were considered forms of financial repression.¹² Among countries with borrowing privileges between 1996 and 2015, 48% also had a closed capital account and 80% had a dual exchange rate regime during the 1980s. By contrast, among countries without borrowing privileges, only 31% had a closed capital account and 55% had a dual exchange rate regime during the 1980s. These patterns suggest that countries turned to borrowing privileges when they lost access to more common forms of financial repression, which were rendered more difficult by increasing levels of capital mobility.

Fig. 4

Share of countries that had dual exchange rates (light grey) or closed capital account (dark grey) in the 1980s among countries without borrowing privileges (left columns) and with borrowing privileges (right columns) during the sample period (1996-2015)

Fig. 5

Share of countries that loosened (light grey) or tightened prudential regulation (dark grey) among countries without borrowing privileges (left two columns) and with borrowing privileges (right two columns) during the sample period (1996-2015)

In Fig. 5, we compare the tightening and loosening of prudential regulation, based on data from the IMF Integrated Macroprudential Policy Database (iMaPP), to the incidence of borrowing privileges. The iMaPP Database codes prudential regulatory policies for a large number of countries and years. It is one of the most comprehensive datasets for the prudential regulatory environment across countries. Figure 5 shows that countries with borrowing privileges in place tighten prudential regulation at about the same rate as countries without borrowing privileges in place (13% compared to 10%); they are more likely to relax prudential regulation (31% compared to 21%). This suggests that borrowing privileges are unlikely to coincide with, or reflect, a broader move toward prudential regulation.

2.2 Fiscal pressure and borrowing privileges

We focus on three indicators of fiscal pressure: borrowing costs for government debt; trade liberalization, which results in a loss of trade taxes; and the need for fiscal autonomy that arises from fixed exchange rates.

For our main measure of borrowing costs, we rely on data on government bond yields (specifically, yield to maturity). Bond yields offer an immediate measure of the demand for government bonds, and therefore provide a close match to our focus on conditions when governments want to increase demand for their debt: because bond coupons are usually fixed at the time of the bond issue, rising bond yields imply falling bond prices, reflecting a decline in the demand for bonds.

For emerging market economies, we use data from the J.P. Morgan EMBI Global Index (Morgan, 2019). The index tracks yields on external debt instruments issued by emerging market governments. It provides a weighted average of yields, with weights determined by the maturity structure of a country’s debt portfolio. The average yield thus closely approximates a country’s bond portfolio. The variable captures the yield demanded by international markets to hold government debt. It is one of the most widely used measures of the performance and composition of emerging market debt,¹³ and the aggregate index is frequently considered a benchmark for sovereign risk (McGuire & Schrijvers, 2003). Data on bond yields from the J.P. Morgan EMBI Global Index are available for 58 countries. We use these bond yields where available.

For countries where the EMBI Global Index does not report yields – including most OECD countries – we obtain data on the average yields of government bonds, weighted again to match the maturity structure of each government’s debt portfolio, from Datastream Refinitiv. We are able to locate data on the average yields of government bonds for 22 additional countries. Combining these data with the EMBI Global Index yields an overall sample of 80 countries. For each country, we only rely on one source (EMBI Global Index yields or average government bond yields). That our results are driven by within-country variation, as described below, rules out that differences in the data source across countries bias our results.

Yields in our sample range from -0.7% for Switzerland in 2015 to over 43% for Argentina in 2002, with an average yield of 6.6%. The within-country variance in our sample slightly exceeds the cross-sectional variance. We transform the variable using the inverse hyperbolic sine transformation, which is almost identical to the log but allows for negative and zero values.

For our second main variable we obtain a country’s average applied tariff rate from the World Bank. Because we are interested in the ability of governments to use tariffs for revenue generation, not for protection, we use the tariff rate weighted by imports. For European Union members, we use the common external tariff. The data are available for 86 countries; the only missing country is Iraq. The trade-weighted average tariff rate ranges from 0% to 40.7%, with an average of 5.6%. Reflecting the wide-spread liberalization of international trade during our sample period, the variable has within-country variance that is almost on par with its cross-country variance. As with yields, we transform the data using the inverse hyperbolic sine.

For the third variable, we use data on exchange rates from Ilzetzki et al. (2019). We consider as fixed exchange rates de facto pegs, pre-announced and de facto crawling pegs, and pre-announced and de facto crawling bands of up to 2%. Following this definition, about two thirds of the observations in our sample have fixed exchange rates. 36 countries in the sample switch their exchange rate regime at least once throughout the sample period. The data on exchange rate regimes are available for all countries and years in our dataset, leaving us with the full sample of 87 countries across 20 years.

We include a small set of control variables. First, we include GDP per capita and log GDP to account for country wealth and market size (data from the World Bank). Larger and wealthier markets could prove more attractive for governments to tap into. They also provide governments with a larger tax base, which increases the government’s ability to repay debt and, in turn, reduces borrowing costs. Countries with larger and wealthier markets are also on average less dependent on trade and less likely to have fixed exchange rates.

Second, democracies are associated with better access to domestic and international debt markets, lower tariffs, and exchange rate regime choices. Democracies may also have different regulatory priorities and be more likely to codify informal practices in law. We use polity scores (Marshall et al., 2017), coding a country as democratic when it reaches a polity score of at least seven. For four countries in our dataset, polity scores are not available. For these, we use the updated democracy indicator from Cheibub et al. (2010). For a more thorough discussion of the link between democratic institutions and borrowing privileges, see Betz and Pond (2023a). We also include a dummy variable indicating membership in the European Union. European Union law all but rules out preferential treatment of a government’s own debt, and membership may reduce borrowing costs. All models also include a cubic year trend. Finally, we control for country-specific effects in our empirical models, thus relying on within-country variation. We consider additional control variables in robustness checks below; we also present models without any control variables in the appendix.

Our dataset comprises a time series cross section with a binary dependent variable. This combination poses special challenges when accounting for country-specific effects. The standard fixed effects model rules out that unobserved differences across countries bias the results. However, unlike in linear models, fixed effects logit models do not allow for the inclusion of countries with no change on the dependent variable. Cook et al. (2020) identify and demonstrate the bias that results from dropping such no-event units in the frequently applied conditional fixed effects logit model.

To address unobserved country heterogeneity, we rely on two estimators that allow for the inclusion of country-specific effects and therefore produce estimates based on differences within countries.¹⁴ First, we follow Mundlak (1978) and Chamberlain (1980), who proposed a class of models to account for unobserved heterogeneity. The model we estimate (i) includes the country-specific mean of each independent variable to account for country heterogeneity and (ii) uses the country mean-deviated value for each independent variable, while (iii) accounting for country-level correlation in the error terms through random effects.

Second, following Cook et al. (2020), we rely on the penalized maximum likelihood estimator proposed by Firth (1993). Cook et al. (2020) demonstrate the usefulness of this estimator in the context of time-series cross-sectional data with a binary outcome, unit fixed effects, and units that do not experience changes on the outcome. The penalized maximum likelihood estimator proposed by Firth (1993) adjusts the likelihood function, which allows for the inclusion of country fixed effects and maintains the full sample. We include a fixed effect for every country that adopted borrowing privileges during the sample period, and a ‘no-event’ fixed effect for all others.

Our research design does not isolate an exogenous source of variation in fiscal pressure, making it difficult to draw causal claims from our results. Indeed, borrowing costs are at least in part a symptom of broader underlying issues, to which borrowing privileges might also respond. Borrowing costs are a plausible indicator of tightened access to credit and therefore fiscal need, but not necessarily the underlying cause. We can, therefore, at best evaluate whether tightened access to credit – for whatever underlying reason – correlates with governments reaching for borrowing privileges.

Similarly, trade liberalization and the exchange rate regime are themselves policy choices, which reflect how policy-makers respond to different trade-offs. They are unlikely to be exogenous to borrowing privileges, which might be a response to those trade-offs as well, and bundled into a larger policy context. This makes it impossible with our research design to credibly identify the causal effect of, for example, trade liberalization. While we seek to address endogeneity concerns from three sources, as described in the following, our results therefore cannot be interpreted as causal effects. Instead, our results are, as correlations, suggestive of the mechanisms we seek to assess.

First, as described above, in addition to various control variables, our models account for country fixed effects, such that our results are driven by within-country variance (in the appendix, we also show that cross-country differences do not explain the results). We thus account for country-specific confounders.

Second, different models of economic governance shape policy choices and economic outcomes, and thus might drive an association between borrowing privileges and our core predictors. Yet, while all three of our core predictors reflect fiscal pressure, they map onto economic governance in different ways. Higher bond yields, trade protection, and fixed exchange rates are plausibly more common in countries that flout liberal economic orthodoxy. Thus, while our measures of fiscal pressure all map onto economic governance in the same direction,¹⁵ we expect their effects to cut in different directions: higher bond yields should be positively associated with borrowing privileges, while higher tariffs indicate less fiscal pressure and should be negatively associated with borrowing privileges; fixed exchange rates increase fiscal pressure and should be positively associated with borrowing privileges. Taken together, our results cannot be explained by economic governance but are instead consistent with fiscal pressure as a motivating force for borrowing privileges.

Third, the extant literature suggests few immediately obvious concerns about reverse causality. Financial development may shape comparative advantage in specific industries (Kletzer & Bardhan, 1987; Beck, 2002) and is closely linked to trade patterns (Rajan & Zingales, 2003; Do & Levchenko, 2007). Trade agreements also often incorporate stipulations about financial services (Dobson, 2007). Yet, to our knowledge, no literature identifies a causal link from policies similar to borrowing privileges to trade openness. Likewise, financial regulatory policies and exchange rate regime choices are closely linked. Indeed, banking and exchange rate crises frequently coincide (Kaminsky & Reinhart, 1999); currency mismatches in bank’s balance sheets amplify the consequences of sudden exchange rate movements (Gabaix & Maggiori, 2015); and foreign-currency holdings as well as policies regulating foreign-currency holdings correlate with exchange rate regime choices (Calvo & Reinhart, 2002). Still, to our knowledge, no literature identifies a causal link from policies similar to borrowing privileges to exchange rate regime choices.¹⁶

Table 1 Fiscal pressure and borrowing privileges

Moreover, a government’s choice of trade policy and exchange rate regimes is a choice of enormous economic implications and of immediate political salience. This contrasts with borrowing privileges. These are obfuscated policies that, absent a government’s default on its debt obligations, are unlikely to gain the attention of ordinary citizens. This alleviates concerns that our research design recovers effects based on reverse causality: a government’s exchange rate regime and trade policy choices, as ‘first-order’ considerations of economic policy-making, are plausibly not driven by the presence of borrowing privileges.¹⁷

2.2.1 Results

Table 1 presents our main results. Odd columns report results from the Mundlak-Chamberlain random effects model; even columns report results from Firth’s penalized maximum likelihood fixed effects model. The first two models present the results for borrowing costs. The next two models present the results for tariff rates. The final two models present the results for the exchange rate regime. The coefficients on our main predictors have the expected signs and are statistically significant in all six models (Fig. 6).

The size of the effects is also substantively significant, as illustrated in Fig. 6.¹⁸ The figure displays the predicted probability of borrowing privileges being in place for the three main predictors, for the random effects models in grey and the fixed effects models in black. For each model, the predicted probability is calculated with the respective predictor at one (within-country) standard deviation below the mean and one standard deviation above the mean. We hold all other variables at observed values and average over observations in the sample, to calculate average marginal effects. The figure also displays 95% confidence intervals for the predicted probabilities.

Fig. 6

Predicted probability and 95% confidence interval at one standard deviation below and above the mean for borrowing costs, trade openness, and exchange rate regime. Grey: Mundlak-Chamberlain logit, random effects. Black: Firth logit, fixed effects. Based on Table 1. Standard deviation is the average within-country standard deviation

As borrowing costs increase from one standard deviation below the mean to one standard deviation above the mean, the probability of borrowing privileges being in place increases by 30%, from 11.5% to 15.1% (all numbers based on the Mundlak-Chamberlain random effects model). Reducing tariffs from one standard deviation above the mean to one standard deviation below the mean increases the probability of borrowing privileges being in place by 40%, from 9.4% to 13.3%. And moving from a floating to a fixed exchange rate doubles the probability of a government implementing borrowing privileges, from 6.9% to 14.0%.

The results provide broad support for our argument. As fiscal pressures on governments increase – captured by higher borrowing costs on financial markets, trade liberalization, and exchange rate constraints that increase reliance on fiscal policy – governments are more likely to tap domestic capital markets for revenue generation. They implement policies that favor government debt over other financial assets, thereby shoring up demand.

Democracies appear to be more likely to implement borrowing privileges. This is consistent with two findings from the extant literature. First, democracies have heightened fiscal needs and face difficulties raising revenue from direct taxation (Bastiaens & Rudra, 2018), increasing the incentive to tap into financial markets as a source of credit. Second, the democratic political process leads governments to reach for less transparent policy measures (Kono, 2006), of which borrowing privileges are one example. At the same time, typical forms of financial repression are less common in democracies, reinforcing the interpretation that borrowing privileges are a distinct policy category. In related work, Betz and Pond (2023a), we elaborate on these patterns and the mechanisms that link democratic institutions to the adoption of borrowing privileges.

2.2.2 Measurement of fiscal need

The results are robust to several modifications, which are reported in the appendix.

Inclusion of all fiscal need components. Our three measures reflect the same underlying issue: fiscal need. In the appendix, we include all three variables simultaneously to assess if the variables differ in their importance and to account for the relationships among these three variables. If, for example, a country selects a fixed exchange rate, it cedes control over monetary policy (and the interest rate) to the anchor country. When including all three variables, all coefficients retain their signs and statistical significance. The coefficients on borrowing costs and fixed exchange rates are substantively unchanged. The coefficient on tariff rates, although still negative and statistically significant, becomes almost 50 percent smaller in magnitude. Further analysis suggests that this loss in magnitude is partially due to the reduction in sample size from including all three variables simultaneously and partially due to the correlation between the independent variables.

Index of fiscal need. Rather than including the three measures of fiscal need separately or independently, we create a composite measure based on the three variables. We create two different versions, one based on the values of each variable (we normalize all variables to range from 0 to 1, such that each contributes equally to the index) and one based on ranks. We then re-estimate the models. As we report in the appendix, both indices are associated with the adoption of borrowing privileges. The results also remain for arbitrary combinations of weights in creating the index, as is shown in the appendix.

Alternative measures of borrowing costs. We measured borrowing costs with bond yields. Our results are robust to using several alternatives. First, we create a composite measure of bond yields to increase our sample coverage. For countries for which our main variable (bond yields weighted by maturity) is not available, we use the unweighted average of one-, five-, and ten-year bond yields (data from Datastream Refinitiv). For countries for which these data are not available either, we use data from the IMF on interest rates on long-term government bonds. For each country, we only rely on one of these measures. This composite measure allows us to extend the data to 86 countries; only Zimbabwe does not report any of these measures.

Second, our results are robust to using only short-term interest rates, neglecting the maturity structure of each country’s debt portfolio, and the average maturity of newly issued debt (from the World Bank). Governments that face temporarily higher costs of accessing financial markets may turn to shorter-term maturities to secure better financing terms. Shorter maturities may reflect rising fiscal pressure.

Third, we use implied default probabilities, as perceived by financial markets, calculated from credit default swap spreads. Credit default swaps allow creditors to purchase insurance from third parties against a borrower’s default. The spreads on credit default swaps indicate the cost of this insurance, which allows backing out the implied default probability. We obtain data on credit default swaps, and the implied default probability for government bonds, for 67 countries from Datastream Refinitiv. Data on credit default swaps is only available for a relatively small number of years, from 2009 to 2015, and we therefore estimate pooled models.

Finally, in our main results, we used yields from the J.P. Morgan EMBI Global Index for countries for which these were available. These are based on international, foreign-currency denominated debt, which matches our focus on constraints from international markets. But they may also limit the comparability across countries, because governments increasingly issue debt in local currency on international markets (Ballard-Rosa et al., 2022), and because many governments issue debt domestically (Claessens et al., 2003). We therefore present additional results using average bond yields, obtained from Datastream Refinitiv. The data are available for fewer countries (our sample size drops to 48), but the results remain robust.

Alternative measures of tariff revenue constraints. Instead of the average tariff rate, we consider several other measures. These results provide a closer link between revenue and borrowing privileges, which helps alleviate concerns that tariff reductions could for example increase revenue if they lead to more efficiency in regulation and growth in the economy.

First, we verify that our main results are robust to using tariff revenue rather than tariff rates. Tariff revenue may more closely reflect fiscal constraints. We opt for tariff rates in the main analysis, as rates are correlated with revenue and have better data coverage in our sample. We replace the average tariff rate with customs revenue as a percent of GDP from the World Bank World Development Indicators. Second, membership in the multilateral trade regime reduces the ability of governments to rely on trade taxes for revenue generation. We replace the average tariff rate with an indicator of whether a country is a member of the World Trade Organization. This variable also alleviates some concerns about endogeneity, as WTO accession is unlikely to be caused by the presence of borrowing privileges. Third, we consider a government’s statistical capacity, using a measure from the World Bank. Countries with lower capacity may have more difficulty raising revenue through domestic taxation, which increases the need for other revenue sources. Fourth, we consider a country’s tax revenue as a percent of GDP. Countries that are less able to raise tax revenue directly should be more likely to tap into domestic financial markets as a revenue source. Finally, we consider natural resource wealth (natural resource production as a percent of GDP). Governments whose economies rely on natural resource extraction can draw on a substantial income source, which lessens the need for borrowing privileges.

Fixed exchange rates in context. Recent research suggests that globalization may actually create a dilemma, as monetary autonomy cannot be sustained without capital controls – even with a floating exchange rate (Rey, 2015). In the Appendix, we introduce a control for capital account openness. Openness is negatively associated with borrowing privileges (although the magnitude of the effect is small), perhaps because governments may suspend openness and implement borrowing privileges when they face fiscal pressure. Nevertheless, the coefficient on the fixed exchange rate remains largely unchanged. We also omit borrowing privileges that pertain specifically to access to foreign exchange, and show that the results are robust to using only borrowing privileges that are targeted at institutional investors and banks.

2.2.3 Robustness checks

In the appendix, we report several models to assess the robustness of our results – including using different control variables, empirical specifications, and sampling and coding decisions. Although our empirical strategy cannot identify causal estimates, we make several attempts to control for the main confounding factors. We present results when controlling for central bank independence, bureaucratic capacity, governance effectiveness, the electoral rule, and the system of government; when including terms-of-trade shocks, political protest, and impending elections to account for political and economic shocks that could drive up borrowing yields; when using a de facto measure of exchange rate stability; and when estimating random effects logit models and conditional fixed effects logit models.

We also report that the results are robust to including controls for the market and regulatory context around borrowing privileges. To capture market conditions, we control for stock market capitalization and turnover, domestic credit to the financial sector, and international debt securities. To capture the regulatory environment, we include controls for capital account openness, the KOF index of globalization, dual exchange rates, prudential regulation, and whether a country is under an IMF loan program in the current or the past year.

We also report the results for different sampling decisions. Borrowing privileges are most common in middle income countries. We therefore report results when limiting the sample to OECD countries, to non-OECD countries, and to middle income countries.¹⁹ The results are robust when omitting OECD countries or examining only middle income countries. The coefficients frequently lose significance when considering only OECD countries. These results are consistent with the attractiveness of imposing borrowing privileges when markets are sufficiently deep that they provide a meaningful source of revenue, but they are mostly unnecessary in the OECD where there is already substantial demand for government debt.

For consistency, we only relied on the IMF AREAER Reports to code borrowing privileges. The reports could be biased if, for instance, they provide more details for countries with fiscal difficulties. Country fixed effects alleviate those concerns only if these country-specific factors remain constant over time. We confirmed the coding of countries without borrowing privileges using a Google search. We tasked a research assistant with surveying the financial regulations for all countries that never had borrowing privileges in place during our sample period.²⁰ We obtained additional borrowing privileges for four countries: Ethiopia, Austria, China, and Sri Lanka. The results remain substantively unchanged when adding these to the original coding.

2.2.4 Borrowing privileges and prudential regulation

Fiscal pressure, and especially fiscal pressure emanating from international markets, can also lead governments to adopt prudential regulation. To differentiate fiscal from prudential motivations as drivers of borrowing privileges, we pursue two strategies.

First, we control for the main motivations to adopt borrowing privileges for prudential reasons. Where domestic financial markets are underdeveloped, high-quality assets might not be available, leading governments to give preference to their own debt for prudential reasons.²¹ All countries in our sample have at least somewhat developed financial markets: all countries in our sample have domestic banks and have issued bonds; only Belize and Ethiopia do not have their own stock markets, and the results are robust to excluding these two countries.

To account for differences in the access to alternative reserve assets, we include controls for the development and liquidity of domestic financial markets: stock market capitalization, stock market turnover, and domestic credit to the private sector, to capture the availability of domestic reserve assets, as well as outstanding total international debt securities to capture the availability of foreign assets (all data from the World Bank Global Financial Development Database). The results, reported in the appendix, are robust to the addition of these variables.

Second, in the appendix we replicate the main models from Table 1, replacing the dependent variable with the measure of prudential regulation from the Integrated Macroprudential Policy Database (iMaPP) by the IMF. We estimate linear regression models. None of the coefficients are statistically significant at the 5% level. Moreover, the coefficients point in the opposite direction from the previous results, indicating that our measures of fiscal pressure lead to less prudential regulation. We also replicate the models for the 24 subcomponents from iMaPP and find only occasionally statistically significant effects; none of the variables has a statistically significant association with all three measures of fiscal need. Taken together, the results do not provide evidence of a move toward prudential regulations when fiscal pressure increases, and they strengthen our interpretation that in times of heightened fiscal pressure, governments turn to domestic financial markets as a source of financing, implementing borrowing privileges to do so.

3 Conclusion

Governments use borrowing privileges to tilt the domestic credit market in their favor. Assembling the first systematic data set of these policies, we show that governments increasingly create a privileged position for their own debt relative to other assets when access to credit becomes tighter, and when trade liberalization and fixed exchange rates increase the importance of fiscal space. Governments thus retain the ability to regulate even seemingly mobile financial assets in the context of global markets. Financial service providers in particular are subject to regulation, as their customers are located in the domestic market. Although their assets may be liquid, the underlying business – providing financial services to local customers – is not.

We leave several topics to future research. Most immediately, this includes the effectiveness of borrowing privileges in lowering borrowing costs. Empirically assessing the effects of borrowing privileges is exceedingly difficult, as we expect the privileges to be imposed precisely at those times when governments anticipate high borrowing costs.

We also leave to future work the emergence and diffusion of borrowing privileges across countries and over time. Borrowing privileges present a fruitful area to examine the extent to which international regulatory standards and prevailing ideas shape the content of domestic policy-making. And at least descriptively, it appears that borrowing privileges become more prevalent over time and, potentially, in clusters. Whether that is driven by global changes in the interest rate environment, by learning and diffusion, or by other dynamics altogether remains an open question. More generally, making sense of the motivations for market reforms in contested policy areas – where policy change can at once be predatory or prudential – is an important area for future research (Chwieroth, 2014).

We highlight two implications of our emphasis on how governments privilege their own debt on financial markets. First, North and Weingast (1989) demonstrated how governments secured better borrowing conditions by turning the creditors of the government into the politically representated. Borrowing privileges, to some extent, reverse this idea: in democracies, borrowing privileges turn the politically represented into the creditors of the government. By increasing widespread domestic ownership of government debt through pension funds, banks, and insurance assets, default becomes politically unattractive, increasing credibility of repayment. Similar to early extensions of the franchise (Stasavage, 2011), a stable base of domestic creditors increases the confidence of creditors in the government’s willingness to repay its debt in the future, and this reciprocal effect facilitates the government’s access to financial markets. Moreover, as Queralt (2022) demonstrates, the emergence of state institutions – including institutions needed to levy taxes – can depend enormously on global economic conditions: where international lending is easily available, governments face fewer pressures to develop institutions that raise funds in the domestic market. This suggests fruitful avenues for considering the interactions between representative institutions, global economic conditions, and financial regulation in shaping government borrowing.

Second, we point to distributional conflicts over financial regulation that are likely to become more salient as financial markets gain prominence in the social, economic, and political life of citizens: between governments and other borrowers; and between governments and their creditors. The first resembles the familiar crowding-out effect, but is reinforced here because governments not only borrow on financial markets, but also tilt them in their own favor. The second distributional conflict, likewise, plays out in the shadow of the government’s ability to turn citizens into creditors, and it coincides with information asymmetries: many citizens may not realize they are creditors of their own government. Our paper thus raises new questions for democratic governance. Understanding how governments resolve the distributional conflicts over financing their expenses, and the relationship between growing financial markets, government revenues, and state autonomy remain important areas for future research.