A. The CBI index

The approach to central bank independence used in this historical tour is based on Cukierman (1992). Cukierman’s notion of central bank independence considers four key elements, which lay the foundations for monetary policy formulation and implementation: (i) the legal mandate (policy objective); (ii) its governing structure (central bank governance); (iii) the powers to control monetary policy vis-à-vis the government (policy formulation); and (iv) the conditions for financing government expenditure (central bank lending). A snapshot of the evolution of these institutional features over the three periods identified above is presented in Table 1.

Based on these four criteria, we construct a CBI index codifying historical central bank legislation covering 17 Latin American countries over a period that spans 100 years. Since the CBI index uses information from central banks’ legislation, it is an index of legal central bank independence exclusively. Our CBI index is a modified version of the widely known Cukierman, Webb, and Neyapti (CWN) index, the most common metric used in the literature to measure central bank independence.⁵ The CBI index evaluates the same four criteria in the CWN index and is calculated by taking the weighted average of them: the central bank governance (20 percent); the legal mandate (15 percent); policy formulation (15 percent); and central bank lending (50 percent). These four criteria are in turn broken down into 15 sub-criteria, as spelled out in Annex 1.

^{Table 1:}

Key features of central bank independence over 100 years

Source: Central bank legislation and its reforms and relevant sections of constitutions.

^{Table 1:}

Key features of central bank independence over 100 years

	Mandate/Functions	Governing arrangement	Monetary policy autonomy	Credit to the government
Early years	- Issue the national currency. - Conduct rediscount and discount operations with banks and the general public. - Provide credit to the public sector on a limited basis. - Work as a depositary and fiscal agent. - Receive deposits from banks, the public sector, and the general public. - Provide clearing for payments.	- Board of Directors comprised of shareholders’ appointees (banks), a minority of government representatives, and sometimes from the business associations and the general public. - The President of appointed by the Board of Directors.	- Before the Great Depression, monetary policy restricted by limitations to issue banknotes and, specifically, to provide credit to the government. - In the fallout of the Great Depression, the restrictions to issue currency were increasingly relaxed and the conditions to finance the government expanded.	- Allowed but on limited amounts depending on each country. Central bank loans are capped in a range of 10 to 30 percent of the central bank paid capital and reserves. - These restrictions were eased in the run-up to Great Depression and in the following years. In addition, specific laws were passed to expand central bank credit to the government.
Developmental phase	Conduct monetary, credit, and exchange rate policies with the aim of fostering the appropriate conditions for an orderly and fast development of the economy.	- Board of Directors comprised of representatives from banks, the government, and business associations. The majority of members belonged to the executive branch. - The President of the Board is appointed by the President of the Republic.	- Governments gradually gained increasing influence on monetary and exchange rate policy formulation and implementation. - The Minister of Finance had veto power over monetary policy decisions.	Allowed but on limited amounts. Restrictions varied for each country, up to a proportion of government revenues in a range between 8 and 15 percent in the previous year. Loan maturity and the interest rate charged also varied.
Golden period	- Preserve price stability. - Maintain the stability of the banking system. - Promote the normal functioning of the payments system.	- Board of Directors comprised of technocrats nominated by the government and appointed by the Congress for a tenure longer than the presidential term. - Dismissal of Board members must be approved by Congress.	The Board of Directors enjoyed autonomy to formulate and execute monetary and exchange rate policy, including adjustments to the policy rate.	Restrictions to finance the government are tighter. Loans must be repaid in the same fiscal year and are charged a market rate. In some countries, credit to the government is banned.

Source: Central bank legislation and its reforms and relevant sections of constitutions.

View Table

^{Table 1:}

Key features of central bank independence over 100 years

	Mandate/Functions	Governing arrangement	Monetary policy autonomy	Credit to the government
Early years	- Issue the national currency. - Conduct rediscount and discount operations with banks and the general public. - Provide credit to the public sector on a limited basis. - Work as a depositary and fiscal agent. - Receive deposits from banks, the public sector, and the general public. - Provide clearing for payments.	- Board of Directors comprised of shareholders’ appointees (banks), a minority of government representatives, and sometimes from the business associations and the general public. - The President of appointed by the Board of Directors.	- Before the Great Depression, monetary policy restricted by limitations to issue banknotes and, specifically, to provide credit to the government. - In the fallout of the Great Depression, the restrictions to issue currency were increasingly relaxed and the conditions to finance the government expanded.	- Allowed but on limited amounts depending on each country. Central bank loans are capped in a range of 10 to 30 percent of the central bank paid capital and reserves. - These restrictions were eased in the run-up to Great Depression and in the following years. In addition, specific laws were passed to expand central bank credit to the government.
Developmental phase	Conduct monetary, credit, and exchange rate policies with the aim of fostering the appropriate conditions for an orderly and fast development of the economy.	- Board of Directors comprised of representatives from banks, the government, and business associations. The majority of members belonged to the executive branch. - The President of the Board is appointed by the President of the Republic.	- Governments gradually gained increasing influence on monetary and exchange rate policy formulation and implementation. - The Minister of Finance had veto power over monetary policy decisions.	Allowed but on limited amounts. Restrictions varied for each country, up to a proportion of government revenues in a range between 8 and 15 percent in the previous year. Loan maturity and the interest rate charged also varied.
Golden period	- Preserve price stability. - Maintain the stability of the banking system. - Promote the normal functioning of the payments system.	- Board of Directors comprised of technocrats nominated by the government and appointed by the Congress for a tenure longer than the presidential term. - Dismissal of Board members must be approved by Congress.	The Board of Directors enjoyed autonomy to formulate and execute monetary and exchange rate policy, including adjustments to the policy rate.	Restrictions to finance the government are tighter. Loans must be repaid in the same fiscal year and are charged a market rate. In some countries, credit to the government is banned.

Source: Central bank legislation and its reforms and relevant sections of constitutions.

The CBI index introduces modifications to the underlying sub-criteria in the CWN index. Most of the changes are incorporated to the sub-criteria characterizing the governing arrangement of the central bank. The CBI index expands the assessment of the rules for the appointment and dismissal to the entire central bank’s board of directors, instead of focusing exclusively on the central bank governor, which is relevant because the legal powers of central bank governors generally represent only a minority fraction within the central bank board. Furthermore, the legal provisions for the appointment and term of office for the central bank governor differ from those that apply to the rest of the board members in a number of countries in the region. The criterion to evaluate the term of office of the central bank governor—and other members of the board of directors—is also different. The CBI index considers not only the absolute length of their appointment, but also their tenure relative to the government. In particular, the CBI index rewards extended or staggered appointments with respect to the government’s term, as a way to strengthen central bank independence from the political cycle. An additional change emphasizes the restrictions on the executive branch to remove members of the board of directors—and not only the central bank governor. The procedures for dismissal of members of the central bank board, including the central bank governor, receive extra weight, since those legal provisions are crucial to preserve the independence of monetary policy formulation. And unlike the CWN, the CBI index penalizes the presence of executive branch representatives at the central board. Given the importance in a historical context of the analysis, the CBI index also assesses the autonomy of exchange rate policy, penalizing the interference of the government as it could likely limit the conduct of monetary policy in small open economies. Lastly, the CBI index adjusts the sub-criteria underlying the conditions for lending to the government to consider the wide array of provisions found in 100 years of central bank legislation in Latin America.

The CBI index varies in a continuous scale from 0 to 1, such that the closer the index is to 1 the higher is the independence of the central bank and vice versa. It assigns higher marks to those institutional arrangements where the executive branch has no power to exert control over the central bank board. Specifically, central banks are considered to be more independent if the government has limited powers to appoint and dismiss the governor and central bank board members, and when their tenure is longer than that of the executive branch. The intuition behind this assumption is that governments and central banks do not always share common interests in the formulation of monetary policy. Governments are often guided by short-term prospects associated with political cycles, whereas central banks work with long-term horizons in order to achieve policy effectiveness, as monetary policy operates with a lag. On the other hand, having a single or primary objective of price stability is associated with more independence, whereas central banks whose mandate includes development objectives receive a low grade. In addition, central banks must be autonomous to formulate and implement monetary and exchange rate policy in line with their inflation mandate and, therefore, government’s interference reduces the CBI index. And, very importantly, restrictions to grant credit to the government—and, more broadly, to the public sector—are rewarded because they help the central bank to maintain control over monetary policy and, arguably, to keep inflation in check. Th CBI index does not include accountability practices, a key dimension of modern central bank independence, given that this feature became an integral component of the institutional design of monetary policy only in the 1990s, when countries introduced checks and balances on central bank independence.

Putting our CBI index at work allows us to portrait the evolution of central bank independence over the course of 100 years. Figure 1 shows that independence fell shortly after central banks were created in the early 1920s and declined even more during the developmental phase when governments took control of central banks. This trend was reversed in the 1990s with the approval of the new legislation that conferred independence to central banks, although the CBI index has declined in the last 15 years, mainly because Argentina and especially Venezuela restricted the autonomy of monetary policy.

Going granular, Figure 2 uncovers the evolution of the CBI subcomponents. The chart corroborates the notion that the initial deterioration of the CBI index was largely driven by the legal reforms that authorized central banks to print money to finance the government when Latin America was hit by the spillovers of the Great Depression and decided to exit the gold exchange standard to stem the economic collapse and foster its recovery. It also shows that the erosion of central bank independence took hold when central banks became in practice development banks, starting in the mid-1940s, with an overarching mandate of promoting economic development, empowered to finance the government and economic activity , and guided by the rules set by the government for both monetary and exchange rate policy. All sub-components of the CBI index increased markedly in the early 1990s when most countries in Latin America approved new legislation conferring independence to central banks. Assigning central banks the fundamental policy objective of achieving and preserving price stability paved the road for the adoption of inflation targeting by a growing number of countries. An expanded analysis of the historical evolution of central bank independence in Latin America follows below.

View Full Size

100 years of central bank independence

(Average CBI index, 1922 – 2021. All Latin America, except Panama)

Citation: IMF Working Papers 2022, 186; 10.5089/9798400219030.001.A001

Source: Central bank legislation (central bank laws and reforms, and relevant constitutional provisions)

• Download Figure
• Download figure as PowerPoint slide

View Full Size

100 years of central bank independence

(Average CBI index subcomponents, 1922 – 2021. All Latin America, except Panama)

Citation: IMF Working Papers 2022, 186; 10.5089/9798400219030.001.A001

Source: Central bank legislation (central bank laws and reforms, and relevant constitutional provisions)

• Download Figure
• Download figure as PowerPoint slide

B. Central Banks’ Early Years

When the Latin American central banks were created in the 1920s (see Box 1), they enjoyed a high degree of legal independence. Most of these central banks were established following the recommendations of U.S. experts led by Edwin Kemmerer.⁶ The fundamental tenet of Kemmerer’s vision was that central banks should endorse the gold exchange standard as a means of securing internal and external stability of the currencies and to grant political independence to central banks. He used to cite the old proverb that says, “we have gold because we cannot trust governments.”⁷

While central banks did not have an explicit mandate, those that followed Kemmerer’s blueprint were assigned instead specific responsibilities, which primarily included issuing as a monopoly the national currency among others (see Table 1). Control of central banks was vested in a board of directors appointed by their shareholders: the government—although without voting powers— domestic and international private banks, and, in some countries, business associations, labor organizations, and private citizens, thus giving voice to key market participants in the formulation of monetary and credit policies. On the other hand, appointing a diversified membership representing the private sector sought to limit bankers’ influence on policy decisions.

Endorsing the gold exchange standard was the defining policy feature of central banks in the 1920s. Central banks were required to preserve the convertibility of their currencies at a fixed exchange rate while keeping an open capital account to allow capital flows to adjust balance of payment disequilibria. In theory, this implied that central banks were not in control of monetary policy, although this was not necessarily true in practice. For example, adjustments of the rediscount rate by the Central Bank of Chile did not follow adjustments Federal Reserve’s discount rate, contemporaneously and within a three-month window. Moreover, adjustments of these rates featured a negative correlation (see Jácome, forthcoming).⁸ And while monetary policy implementation was restricted because central banks were allowed to issue banknotes only if they were backed by international reserves, mostly gold, banknotes in circulation were initially backed with wide margins, which left central banks with room to expand rediscount operations and provide financial support as lender of last resort as necessary.⁹ Nonetheless, this restriction became binding in the run-up to the Great Depression as capital outflows soared and, hence, international reserves declined. As a result, banknotes in circulation shrunk and interest rates increased.¹⁰

In addition, central banks’ charter crafted the rules for limiting the provision of credit to the government. Central banks could not purchase government securities in excess of a given fraction, ranging from 10 to 30 percent of their paid up capital and reserves. Imposing such restriction was necessary to leave behind the years when note issue was directed to finance the government, which elicited exchange rate instability and inflation. Putting a cap on financing the government and other public sector institutions was also meant to leave room for financing economic activity given the existing limitations that central banks faced to issue money.

The initial architecture of Latin American central banks was quickly shaken by the advent of the Great Depression. As the world economy and global trade collapsed in the late 1920s, Latin American exports plunged and economic activity slowed down and eventually fell into depression.¹² Trade and output decline, in turn, lowered fiscal revenues and, therefore, Latin American countries defaulted on their foreign debt, which severely constrained government external borrowing. Since adherence to the gold exchange standard imposed insurmountable restrictions that impeded central banks to respond countercyclically to the external shock, and given the dire state of public finances, countries abandoned the gold exchange standard and, thus, stopped securing the convertibility of their currencies. Most countries preserved a fixed exchange rate but were free to adjust it as needed, which, together with the introduction of capital controls, gave central banks full monetary policy discretion to pursue expansionary policies aimed at financing the government and helping economies bounce back.

The policies executed to recover from the Great Depression’s economic fallout tested the independence of central banks. In addition to curtailing interest rates, central bank money was required to fund countercyclical fiscal policy at preferential interest rates amid heightened tensions between the government and monetary policy authorities. And because legal restrictions prevented central banks from providing credit to the government in large scale, central bank laws in several countries were amended, or Congress passed specific legislation circumventing the existing restrictions in the central bank charter. As a result, credit to the government surged and became the largest central bank asset by 1934, reaching 70 percent of assets in Chile, 60 percent in Colombia—including loans to the mortgage bank—, and over 40 percent in Ecuador (see Figure 3). Expanding credit to the government also required relaxing the existing restrictions to issue money.¹³

By the mid-1930s, the worst of the crisis in Latin America was over as economic activity started to pick up. New central banks were created, in El Salvador, Argentina, and toward the end of the decade, in Venezuela, with a mandate of regulating the amount of money and credit, given the absence of an international monetary anchor for domestic and international prices. The central bank in Argentina was also responsible for accumulating sufficient international reserves to lean against the wind and moderate the adverse effects of “export and foreign investment volatility,” and importantly, was in charge of regulating and supervising banks. The latter was in contrast to other countries in the region where such responsibilities rested on a separate entity. This institutional arrangement, where monetary policy and banking supervision is conducted under one roof was replicated later when the central banks in Brazil, Paraguay, and Uruguay were created.¹⁴ In addition, the new central banks were vested with more stringent restrictions to finance the government. For example, in Argentina, central bank assets were largely driven by the accumulation of international reserves until 1945—the first signs of fiscal dominance appeared only in 1943 (see Figure 3), when the central bank was required by law to provide credit to the state-owned Banco Nación to finance the purchase of the agriculture harvest of that year.

View Full Size

Central Bank Assets and Credit to the Government

(1930 – 1945, except for Argentina)

Citation: IMF Working Papers 2022, 186; 10.5089/9798400219030.001.A001

Source: Argentina: Central Bank of the Republic of Argentina, Annual Reports; Chile: Central Bank of Chile, Annual Reports; Colombia: Revista del Banco de la República 1930 – 1935 and Informe del Gerente a la Junta Directiva del Banco de la República 1936 – 1945; Ecuador: Banco Central del Ecuador (1997).

• Download Figure
• Download figure as PowerPoint slide

However, central bank independence was a casualty of the recovery. While monetary policy succeeded in pulling Latin American economies from a cliff, this came at the cost of undermining central banks’ independence for decades to come. Political pressure on central banks subsided as economies recovered and public finances were in stronger footing, but the reforms introduced to central bank legislation in the early-1930s remained in place.

All in all, central bank independence declined in Latin America. Because of the reforms to central bank legislation introduced during the Great Depression and its aftermath, and despite the new central banks were born with high independence (see Figures 2A.1 and 2A.2 in Annex 2), the median CBI index declined—from 0.73 to 0.66 between 1930 and 1945. And given that there were two groups of central banks featuring low and high levels of independence, the dispersion of the CBI index—measured by the standard deviation—almost doubled, from 0.044 to 0.081 from 1930 to 1945.

Monetary policies implemented in Latin America to cope with the adverse effects of the Great Depression and their aftermath yield an important lesson. There is no doubt that, in absence of fiscal space, amending legislation to expand central banks’ balance sheets—mainly to extend credit to the government—was a necessary step to prevent a larger collapse of output and to bolster economic recovery. However, financing the government saw the seeds of future fiscal dominance because those extraordinary measures were not reversed once the economies bounced back. This allowed governments to continue resorting to central bank financing, knowing that they would get funding at subsidized financial conditions without paying the political cost of raising taxes and/or cutting expenditures. In practice, central bank credit at subsidized rates became a key source of government financing for several decades.

C. The Developmental Phase

Any vestige of central bank independence eventually disappeared, in some countries more than in others, from the mid-1940s onward. In the context of a new international monetary order defined by the Bretton Woods agreement, and the adoption of an inward-oriented development strategy by the largest Latin American countries, a new central banking paradigm emerged, one in which monetary policy was subservient of government prescriptions. The new paradigm meant in practice that central banks turned into some form of development banks.

Most countries endorsed the Bretton Woods agreement in the second half of the 1940s, which required them to maintain fixed, although adjustable, exchange rates and to commit to the convertibility of their currencies against the U.S. dollar. Nonetheless, central banks could maintain control of monetary policy because of the enhancements of capital controls introduced in the 1930s. At the same time, several countries, in particular the Southern economies, deepened the policies—introduced in the 1930s—that promoted import substitution industrialization as an alternative development strategy.¹⁵ The Central American countries, though, did not engage on import substitution industrialization and instead continued exporting agriculture commodities.

The new development strategy engulfed monetary policy. Because there was a sense that the existing monetary legislation was obsolete—as it was enacted to secure the convertibility of the currencies into gold—major reforms to the central banks’ law were approved.¹⁶ A first critical reform was to expand central banks’ mandate to bring on board the objective of fostering economic development. While some of the new laws also made references to inflation, for example, requiring central banks to “prevent inflationary and deflationary trends,” as in Chile and Ecuador, or “preserve the internal value of the currency” as in Peru and Venezuela, promoting economic development became the overriding objective of monetary policy.

To enforce the new policy objective and their developmental role, the new laws redefined the central banks’ government arrangement. Legal changes included a larger presence of government representatives at the central banks’ board of directors than in the past, although the extent of the government’s influence at the central bank board varied across countries and over time.¹⁷ As a result, governments enhanced their influence on the formulation of monetary policy and even on its implementation. Moreover, they shaped policy decisions even from outside the central bank— at different periods of time.¹⁸

The monetary policy reforms also vested central banks with powers to expand the financing of fiscal expenditure, thus inflicting a lasting damage to central banks’ independence. In an environment of extended financial repression that made unattractive for commercial banks to finance the government, central banks were called upon to fund persistent fiscal deficits that resulted from expanded government expenditure and subsidies. Like in the 1930s, this financing materialized either because new central bank laws were promulgated or the existing ones reformed, or because specific legislation was passed to circumvent the central bank charter.¹⁹ Credit operations were typically provided at preferential interest rates and especially at negative rates, measured in real terms. In addition, ad hoc legislation was frequently passed to restructure governments’ debts with the central bank, extending its maturity and reducing the interest rate paid for those obligations. Another form of central bank financing to the government—albeit more difficult to measure—was the provision of credit to state-owned financial institutions, which, in turn, extended credit to the central government, local governments, and other public sector entities. Therefore, fiscal dominance was, to a great extent, a defining feature of monetary policy, which responded to persistent fiscal deficits across the region, as extensively analyzed for eleven Latin American economies starting in 1960 by Kehoe and Nicolini (2022).

^{Table 2:}

Central bank balance sheets in Latin America

(Index numbers. Selected countries. 1945 – 1970)

Sources: For Argentina and Chile: Central banks’ annual reports. For Colombia, Revista del Banco de la República, several issues. For Ecuador: Banco Central del Ecuador (1997)

^{Table 2:}

Central bank balance sheets in Latin America

(Index numbers. Selected countries. 1945 – 1970)

	1945	1950	1955	1960	1965	1970
Argentina	100	677	1,648	3,416	8,323	19,292
Chile	100	314	1,984	13,577	77,316	420,076
Colombia	100	183	190	509	1,519	2,418
Ecuador	100	159	220	330	466	1,396

Sources: For Argentina and Chile: Central banks’ annual reports. For Colombia, Revista del Banco de la República, several issues. For Ecuador: Banco Central del Ecuador (1997)

View Table

^{Table 2:}

Central bank balance sheets in Latin America

(Index numbers. Selected countries. 1945 – 1970)

	1945	1950	1955	1960	1965	1970
Argentina	100	677	1,648	3,416	8,323	19,292
Chile	100	314	1,984	13,577	77,316	420,076
Colombia	100	183	190	509	1,519	2,418
Ecuador	100	159	220	330	466	1,396

Sources: For Argentina and Chile: Central banks’ annual reports. For Colombia, Revista del Banco de la República, several issues. For Ecuador: Banco Central del Ecuador (1997)

In countries where credit to the government and commercial banks reached sizable amounts, central banks could hardly manage their balance sheet. Because there was a chronic systemic liquidity surplus, it was difficult for the central bank to drain enough money from the market of bank reserves to keep monetary aggregates in check. This happened in Argentina and especially in Chile, where the central bank balance sheet exploded during 1945 to 1970, while in other countries, like Colombia and Ecuador, the pace of increase was slower (see Table 2). Such asymmetry was reflected on inflation performance as the average annual rate surpassed 30 percent in Chile and was close to 25 percent in Argentina in that period, whereas in Colombia inflation was on average 10 percent and only 5 percent in Ecuador.

Except in the Southern Cone countries, following the demise of the Bretton Woods arrangement in the early 1970s, monetary policy started to play a secondary role. Latin America, in general, initially benefitted from a commodity boom that released pressure on central bank financing to the government. Immediately after, capital inflows—that surged on the back of the 1972–1974 oil shock—poured into medium-size and large economies, thus boosting major domestic credit expansions and financing growing fiscal and external disequilibrium. Chile and Argentina followed a different path. In Chile during 1970 to 1973 and in Argentina between 1973 and 1976, where left wing political parties rose to power, monetary policy was a critical component of the governments’ policy agenda. With monetary policy playing a subservient role to the government, the central bank balance sheet surged such that the increase in assets reached close to 40 percent of GDP in Argentina in 1976 and 80 percent in 1973 in Chile, in both cases in response to the increase in credit to the government—and in Argentina also because the central bank absorbed credits and deposits from commercial banks. Such an expansionary monetary policy took its toll on inflation as they hit three-digit rates. Nonetheless, “macroeconomic populism,” as characterized by Dornbusch and Edwards (1991), did not last much. A political coup took place in both countries bringing to power governments from the opposite end of the political spectrum, which shifted the emphasis of economic policies to stabilizing the economy and restoring market conditions to guide price formation. And while governments continued defining the direction of monetary policy, pressures to lend to the government receded. Anti-inflation efforts relied on a forward-looking exchange rate anchor, known as “tablita,” to drive inflation expectations. In other countries, like in Colombia and Ecuador, the amount of credit to the government was negligible. However, monetary policy was still expansionary as much of Latin America received cross-border capital flows that fueled a bank credit boom. Fiscal expenditure was also partially funded with foreign debt.

The persistent increase in aggregate demand financed with capital inflows proved to be unsustainable. In the early 1980s, a triple crisis (debt, banking, and currency) was triggered by the drastic increase in interest rates in the U.S. and the U.K. Monetary policy then became accommodative to prevent the meltdown of financial systems and support quasi-fiscal responsibilities. Thus, in the rest of the decade, most central banks were subject to a combination of fiscal and financial dominance that led to an explosive growth of their assets, sometimes also associated with increasing operational losses, which became another source of monetization that hampered central banks’ capacity to handle monetary policy. Operational losses were mostly a legacy of the monetization of the banking crises and, in some cases, of distortive interest rate and exchange rate policies.²⁰

Because the government influence on monetary policy took hold throughout the developmental phase, the median CBI index declined to 0.54 in 1990 (see Figures 2A.3 and 2A.4 in Annex 2), while the standard deviation kept growing. At the country level, by 1990, the level of central bank independence had reached record lows, with the exception of Chile that in 1989 passed a new law of the central bank that pioneered monetary policy independence in Latin America.

D. The Golden Period

The 1990s marked a turning point for central bank independence in Latin America. Because macroeconomic instability and stagnation in the previous decade had taken a high toll on living conditions and was a major source of social discontent, governments decided to undertake a major structural reform that conferred political independence to central banks. Thus, the vast majority of countries in Latin America approved new central bank legislation or passed major reforms to the central bank charter.²¹ The key tenet of central banks’ institutional reform was to prevent politically motivated short-sighted influences on monetary policy from happening and require central banks to focus on defeating inflation.²² Independence was enshrined in the central bank charter and, in some countries—like in Chile, Colombia, and Mexico—, in the constitution, with the view of making it difficult to reverse this reform. Countries also stopped issuing parallel laws to circumvent central bank independence. In exchange for granting independence, in most cases, legislation held central banks accountable.

Specifically, the reform endowed most central banks with an unequivocal mandate of pursuing price stability. Most Latin American central banks also became independent from governments. Central banks’ boards of directors, including the governor, became appointed for a tenure that exceeded—or overlapped—with the country’s presidential constitutional term, with the view of preventing the subordination of monetary policy to the countries’ political cycle. The law also required in many countries that Congress confirm board members following executive branch nominations. In addition, unlike in the past, the central bank board rarely included private sector representatives. A key feature of political independence was to remove members of the government from the board of directors, although, in some countries, the Minister of Finance continued being a member, often without voting powers.²³ And to strengthen political independence, in many countries the new legislation specified strict legal grounds for removing members of the central bank board of directors and put the final decision in the hands of the legislative or the judicial branch, following a well-established due diligence process.

The new governing arrangement empowered central banks to formulate and execute monetary policy without government interference. Starting in the 1990s, most of central banks enjoyed not only instrument independence, but also goal independence.²⁴ Central banks had the authority to set alone their policy goal, typically the target for inflation, except in few countries like Brazil and Colombia, where the Minister of Finance presides the National Monetary Council and the Monetary Board, respectively—these central banks have instrument independence. Enjoying goal and instrument independence was considered necessary in Latin America to restrict the governments’ interference in policy formulation and execution, given the region’s history of high inflation. Yet in some countries the central bank was not vested with the authority to manage autonomously exchange rate policy. In Mexico and Venezuela, the government retained the power to co-participate in the formulation of exchange rate policy, which had the potential for interfering on the central banks’ operational independence. However, the relevance of this restriction diminishes when countries introduce exchange rate flexibility, like in Mexico.

Chief among the monetary policy reforms was to prevent fiscal dominance. In some countries, the new legislation banned direct or indirect central bank financing to the government. In others, the central bank was authorized to provide advances to help the government cope with seasonal liquidity shortfalls. However, this financing was conceded in limited amounts and the government had to pay it back within the same fiscal year. Banning the provision of credit to the government was elevated in some countries to a constitutional level—like in Chie, Guatemala, and Peru—to make its reversion harder. Banning or restricting fiscal deficit financing reflected the consensus that central bank funding to the government has historically been the main cause of the chronic inflation that afflicted the region (Kehoe and Nicolini, 2022).

Holding central banks accountable was a novelty. Until the 1990s, central banks in Latin America simply published an annual report describing the country’s overall economic performance, the policies they adopted in the given year, and their financial statements. However, this publication served essentially as a historical record. Since the reforms of the 1990s accountability became a check and balances institutional routine, given that independence involved delegating important state powers to an institution governed by unelected officials. With the turn of the century, accountability was complemented with enhanced transparency, in particular in countries that adopted inflation targeting. In these countries, central banks not only disclosed their policy objective, but also began to explain how they formulate monetary policy, the content of policy decisions, and the outlook for inflation and output to guide economic agents’ decisions and their inflation expectations. The enhanced transparency associated with the adoption of inflation targeting resulted in a better anchoring of inflation expectations.²⁵ Accountability and transparency also made central banks less likely to behave in a dynamic inconsistent fashion, thus underpinning central banks’ credibility.

Building on the political and operational independence granted to central banks, and with an unequivocal mandate of defeating inflation, monetary policy delivered the expected results in most countries. To achieve this outcome, the support from other structural reforms implemented during the second half of the 1980s and the early 1990s was critical.²⁶ Importantly, the effectiveness of monetary policy would have not been possible without having in place sounder fiscal policies than in the past. In addition, exogenous tailwinds contributed to this successful result, in particular, enhanced globalization that allowed countries to import cheaper goods coming mostly from China.²⁷ By the late 1990s, an increasing number of central banks started to modernize their monetary policy framework by introducing inflation targeting, which helped to attain low and more stable inflation in the last 20 years (see Carrière-Swallow and others, 2016).²⁸ The importance of central bank independence for inflation targeting has been widely recognized in the literature as an institutional mechanism to infuse credibility on inflation targeting and price stability and to eliminate the time inconsistency problem.²⁹

The wave of reforms implemented in the first half of the 1990s increased significantly the independence of central banks. The median CBI index rose from 0.54 to close to 0.82 from 1990 to 2020, despite the fall in the autonomy of monetary policy introduced in a handful of countries, especially in Argentina and Venezuela during the 2010s. Going granular into the criteria that accounts for the different aspects of the CBI index, suggests that the latest reforms introduced improvements in all areas considered in the index. However, on average, major progress came from (i) refocusing the mandate of central banks on fighting inflation as the primary policy objective, and (ii) instituting a governing arrangement dissociated from the executive branch and, hence, with autonomy to formulate monetary policy without paying attention to political cycle considerations (see Table 3).

Data

The econometric analysis relies on country-level yearly data for a sample of 17 Latin American countries for the period 1940-2019, conditional on availability. We build on the long time series of central bank independence, codifying central bank legislation according to the criteria as described in Annex 1. Data on real GDP comes from the Maddison Project (see Bolt and Luiten van Zanden, 2020). Data on inflation and Central Bank Claims on the Central Government come from IFS. For the early part of the sample, data were obtained from scanned records of original IFS hard copies. These sources are complemented with data on exchange rate regimes from Ilzetzki, Reinhart, and Rogoff (2019), data on crises from Laeven and Valencia (2020), and the IMF structural reform database.

Following the literature, our data excludes outliers on both ends of the inflation distribution. In particular, we winsorized values at the 3 percent level (1.5 in each tail of the distribution). Our final dataset contains between 1200 and 1100 observations depending on our econometric specification.

A. Econometric Approach

We study the historical link between central bank independence in Latin America through several empirical approaches. First, we estimate panel regressions in the spirit of Jácome and Vázquez (2008), Acemoglu and others (2008), and Garriga and Rodriguez (2020). The baseline specification takes the following form:

where π_i,t is the inflation rate³⁵ in country i at time t,, α_c is a country fixed-effect, CBI_i,t is the index of central bank independence, and X_i,t–1 captures lagged country-specific variables typically associated with inflation, including the output gap, lagged inflation, the exchange rate regime³⁶, and net claims of the central bank on the central government normalized by the money base.³⁷ The latter variable captures de facto links between the central government and the central bank. This panel approach is aimed at estimating the “average” response of inflation to differences in central bank independence. In addition to the baseline variables, the specification in (1) is expanded by including additional controls, such as global inflation, proxied by U.S. inflation, a banking crisis dummy,³⁸ the country’s exchange rate regime, and, for a shorter time span, the state of structural reforms in the country.

Our second approach focuses on estimating the dynamic cumulative impact of changes in central bank independence on inflation. To do so, we follow the local projection method proposed by Jordà (2005). The method has the advantage that it does not constrain the shape of the impulse response functions and is therefore less sensitive to misspecification than estimates of VAR models (Jordà and Taylor, 2016). The approach is similar to that followed by Carrière-Swallow and others (2021) when studying exchange rate pass-through. The benchmark specification is as follows:

where for each horizon h= {0,1,2,3,4,5} we estimate equation (2). Our coefficient of interest will be β^h, which captures the cumulative dynamic impact of differences in the CBI on cumulative changes in the price level associated with the transformed inflation measure. Notice that, in addition to the controls included in (1) and lags of these variables, we also control for past values of the CBI and inflation. The inclusion of these additional controls aims at taking into account past dynamics of our variables of interest, which may be important determinants of current values.

One additional attractive feature of the local projection method is that it allows to study state contingencies in the response of the dependent variable to shocks in a parsimonious and flexible way. Making use of this feature, in some of the exercises below we study whether the impact of changes in central bank independence on inflation varies with country-level state variables, such as the initial inflation level, and the state of other structural reform areas. When focusing on these exercises, we estimate the following state-dependent local projection:

in which S_i,t stands for a dummy variable that takes value 1 if a specific country characteristic is satisfied in t – 1 (as described above) and zero otherwise. As in the baseline local projections, we cap our horizon at 5 years.

In addition to studying average effects and dynamic results, the paper assesses the role that CBI has in preventing tail events. In particular, we study the extent to which higher levels of central bank independence are successful in reducing the right tail of the inflation distribution and in preventing episodes of high inflation. The first objective is tackled by estimating a panel quantile regression model as in Gelos and others (2022). The quantile regression framework quantifies the impact that CBI has on different quantiles of the inflation distribution. To study the role of central bank independence in reducing the likelihood of high inflation episodes, we estimate a fixed-effects linear probability model, which follows a similar specification as (1), but where the left hand side variable is a dummy variable taking value 1 if inflation crosses a given threshold. For robustness we estimate such model for several thresholds.

A. Evidence from Panel Regressions

Using our historical coverage of 17 Latin American countries, we start by revisiting the panel regression evidence of the link between inflation and CBI (Table 4). Our results show that inflation is negatively associated with higher levels of central bank independence, and positively associated with above-potential output levels and with increases in net central bank claims on the central government, which highlights the importance of fiscal dominance in Latin America (Column 1).

The estimated coefficient for the CBI index is both statistically significant and economically meaningful. The coefficient implies that an increase in the CBI index of 30 points, equivalent to moving from the 25th to the 75th percentile of the historical distribution of the CBI, would result in inflation declining from its historical median value to the 25th percentile of the historical distribution. For context, a 30-point change in the CBI index is similar in magnitude to what was observed in LAC countries when they introduced legislation in the late 1980s/early 1990s to strengthen central bank independence.

The link between CBI and inflation is robust to the inclusion of additional controls. In particular, we find that the magnitude and statistical significance of the CBI coefficient remains virtually unchanged when we sequentially control for international inflation (proxied by the US inflation, Column 2), for the country’s exchange rate regime (Column 3), for a dummy of banking and sovereign crises (Column 4), and for time fixed effects (Column 5), although the latter specification results in a slightly lower and less statistically significant point estimate.

Our results also suggest that the link between CBI and inflation is not affected by the period of analysis. To test the stability of the link between inflation and the central bank independence, we interact the CBI index with two dummies, marking a pre and post period. In column 6 we compare the pre-1990 and post-1990 period, a useful split of the sample as 1990 roughly coincides with the wave of central bank reforms seen in Latin America. In column 7 we split the sample between the pre- and post-1970 period, which boils down to studying the period before and after the US dollar convertibility for gold was cancelled. In both cases we find that the CBI coefficient is statistically significant for both the pre and post period and of similar magnitude.

Next, we study the robustness of our results to the inclusion of a structural reforms index and to alternative estimation methods that tackle certain endogeneity problems. As highlighted by Jácome and Vázquez (2008), structural reforms and central bank independence go hand-in-hand, and our baseline results may be capturing the effects of pro-competition reforms that lower prices and, as a result, inflation. To study the importance of the omission of structural reforms in our baseline regressions, we use the IMF Structural Reform Database, which covers reforms implemented over the period 1973-2014 in four broad areas: i) domestic finance ii) trade iii) product market, and iv) labor market (see IMF (2019)). Given that the country coverage and time frame changes when we include the structural reform index, we begin by estimating a similar specification as the one in column (4), but excluding country/years for which the structural reform index is not available. As shown in column (8), the estimated coefficient for CBI remains negative and significant in the restricted sample. Next, we estimate our baseline equation adding the structural reform index as a control (column (9)). While the point estimate of CBI declines (and the statistical significance falls), we still find a negative association between the CBI index and inflation. Finally, column (10) shows results of the estimation of our baseline model using the Arellano-Bond estimator, which deals with endogeneity issues arising in dynamic panel settings. Results show that the estimated coefficient for CBI remains statistically significant, albeit the magnitude of the coefficient is lower than in column (4).

^{Table 4.}

Central Bank Independence and Inflation

Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1

^{Table 4.}

Central Bank Independence and Inflation

Dependent variable: inflation	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)
Central Bank Independence (CBI) index	-0.147***	-0.139***	-0.121***	-0.121***	-0.108**			-0.188***	-0.0996*	-0.108***
	(0.0337)	(0.0336)	(0.0306)	(0.0302)	(0.0495)			(0.0457)	(0.0484)	(0.0202)
Inflation (t-1)	0.716***	0.711***	0.553***	0.552***	0.565***	0.551***	0.535***	0.501***	0.439***	0.628***
	(0.0356)	(0.0362)	(0.0465)	(0.0460)	(0.0506)	(0.0478)	(0.0455)	(0.0650)	(0.0440)	(0.0308)
Net Central Bank Claims on	0.0162***	0.0162***	0.0246***	0.0245***	0.0328***	0.0245***	0.0269***	0.0314***	0.0362***	-0.0229
Central Government (% of Base Money, t-1)	(0.00411)	(0.00405)	(0.00428)	(0.00459)	(0.00672)	(0.00470)	(0.00464)	(0.00671)	(0.00490)	(0.0171)
Output gap (t-1)	0.598***	0.578***	0.610***	0.611***	0.700***	0.611***	0.598***	0.300	0.246	0.559***
	(0.161)	(0.166)	(0.160)	(0.163)	(0.164)	(0.163)	(0.160)	(0.211)	(0.197)	(0.0930)
Fixed Exchange Rate Dummy (t-1)			-12.10***	-12.10***	-10.65***	-12.09***	-11.52***	-13.82***	-12.47***	-11.39***
			(1.770)	(1.772)	(1.664)	(1.746)	(1.847)	(3.317)	(2.362)	(1.015)
Semi Flexible Exchange Rate Dummy (t-1)			-10.42***	-10.42***	-9.148***	-10.45***	-10.54***	-10.39***	-7.751***	-9.247***
			(1.831)	(1.823)	(1.742)	(1.895)	(1.806)	(2.427)	(1.573)	(1.024)
Flexible Exchange Rate Dummy (t-1)			-9.466***	-9.469***	-8.929***	-9.479***	-9.482***	-11.67***	-9.530***	-8.790***
			(1.773)	(1.774)	(1.632)	(1.793)	(1.768)	(2.253)	(1.812)	(0.985)
Banking Crisis Dummy (t-1)				0.153	-1.015	0.147	-0.167	-2.228	-1.946	1.318
				(1.492)	(1.613)	(1.496)	(1.529)	(2.099)	(1.862)	(0.980)
US Inflation (t-1)		0.130 (0.0988)	0.252** (0.0899)	0.252** (0.0898)		0.253** (0.0905)	0.181 (0.106)	0.0431 (0.185)	-0.369** (0.142)	0.119 (0.0837)
CBI*year<=1990						-0.123*** (0.0385)
CBI*year>1990						-0.125** (0.0484)
CBI*year<=1970							-0.146*** (0.0383)
CBI*year<1970							-0.177*** (0.0470)
Structural Reform Index (t-1)									-9.513*** (2.790)
Inflation (t-2)										-0.0380 (0.0261)
Constant	13.74***	12.78***	22.37***	22.40***	19.95***	22.60***	25.17***	30.46***	19.10***	20.70***
	(2.516)	(2.520)	(3.185)	(3.135)	(5.463)	(4.076)	(3.689)	(5.420)	(6.000)	(1.778)
Country Fixed Effects	1,170	1,170	1,133	1,133	1,133	1,133	1,133	505	504	1,111
Year Fixed Effects	0.675	0.675	0.711	0.711	0.751	0.711	0.715	0.784	0.804
Observations	1,170	1,170	1,133	1,133	1,133	1,133	1,133	505	504	1,111
R-squared	0.675	0.675	0.680	0.680	0.727	0.680	0.683	0.754	0.784
Number of id	17	17	17	17	17	17	17	16	16	17

Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1

View Table

^{Table 4.}

Central Bank Independence and Inflation

Dependent variable: inflation	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)
Central Bank Independence (CBI) index	-0.147***	-0.139***	-0.121***	-0.121***	-0.108**			-0.188***	-0.0996*	-0.108***
	(0.0337)	(0.0336)	(0.0306)	(0.0302)	(0.0495)			(0.0457)	(0.0484)	(0.0202)
Inflation (t-1)	0.716***	0.711***	0.553***	0.552***	0.565***	0.551***	0.535***	0.501***	0.439***	0.628***
	(0.0356)	(0.0362)	(0.0465)	(0.0460)	(0.0506)	(0.0478)	(0.0455)	(0.0650)	(0.0440)	(0.0308)
Net Central Bank Claims on	0.0162***	0.0162***	0.0246***	0.0245***	0.0328***	0.0245***	0.0269***	0.0314***	0.0362***	-0.0229
Central Government (% of Base Money, t-1)	(0.00411)	(0.00405)	(0.00428)	(0.00459)	(0.00672)	(0.00470)	(0.00464)	(0.00671)	(0.00490)	(0.0171)
Output gap (t-1)	0.598***	0.578***	0.610***	0.611***	0.700***	0.611***	0.598***	0.300	0.246	0.559***
	(0.161)	(0.166)	(0.160)	(0.163)	(0.164)	(0.163)	(0.160)	(0.211)	(0.197)	(0.0930)
Fixed Exchange Rate Dummy (t-1)			-12.10***	-12.10***	-10.65***	-12.09***	-11.52***	-13.82***	-12.47***	-11.39***
			(1.770)	(1.772)	(1.664)	(1.746)	(1.847)	(3.317)	(2.362)	(1.015)
Semi Flexible Exchange Rate Dummy (t-1)			-10.42***	-10.42***	-9.148***	-10.45***	-10.54***	-10.39***	-7.751***	-9.247***
			(1.831)	(1.823)	(1.742)	(1.895)	(1.806)	(2.427)	(1.573)	(1.024)
Flexible Exchange Rate Dummy (t-1)			-9.466***	-9.469***	-8.929***	-9.479***	-9.482***	-11.67***	-9.530***	-8.790***
			(1.773)	(1.774)	(1.632)	(1.793)	(1.768)	(2.253)	(1.812)	(0.985)
Banking Crisis Dummy (t-1)				0.153	-1.015	0.147	-0.167	-2.228	-1.946	1.318
				(1.492)	(1.613)	(1.496)	(1.529)	(2.099)	(1.862)	(0.980)
US Inflation (t-1)		0.130 (0.0988)	0.252** (0.0899)	0.252** (0.0898)		0.253** (0.0905)	0.181 (0.106)	0.0431 (0.185)	-0.369** (0.142)	0.119 (0.0837)
CBI*year<=1990						-0.123*** (0.0385)
CBI*year>1990						-0.125** (0.0484)
CBI*year<=1970							-0.146*** (0.0383)
CBI*year<1970							-0.177*** (0.0470)
Structural Reform Index (t-1)									-9.513*** (2.790)
Inflation (t-2)										-0.0380 (0.0261)
Constant	13.74***	12.78***	22.37***	22.40***	19.95***	22.60***	25.17***	30.46***	19.10***	20.70***
	(2.516)	(2.520)	(3.185)	(3.135)	(5.463)	(4.076)	(3.689)	(5.420)	(6.000)	(1.778)
Country Fixed Effects	1,170	1,170	1,133	1,133	1,133	1,133	1,133	505	504	1,111
Year Fixed Effects	0.675	0.675	0.711	0.711	0.751	0.711	0.715	0.784	0.804
Observations	1,170	1,170	1,133	1,133	1,133	1,133	1,133	505	504	1,111
R-squared	0.675	0.675	0.680	0.680	0.727	0.680	0.683	0.754	0.784
Number of id	17	17	17	17	17	17	17	16	16	17

Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1

Table 5 studies the relative importance of different CBI components in explaining inflation. In particular, it shows results of an exercise where we estimate two variations of equation (1): one where we control for each subcomponent of the CBI index, and one where we control for each subcomponent and an aggregate index of other subcomponents.^{39 40} Column 1 shows that a higher value for the central bank governance component, which can be interpreted as central banks being less exposed to political influence, is associated with lower inflation. This result holds even when we control for the index of all other subcomponents (column 2). Higher values of the policy objective and policy formulation components are also associated with lower levels of inflation (columns 3 and 5, respectively), but their statistical significance vanishes when we control for the aggregate index for other subcomponents (columns 4 and 6). The subcomponent capturing the extent to which central banks can lend to the central government appears as an important determinant of inflation. Column 7 shows that higher values of the index (that is, more limits to central bank lending) are associated with lower inflation, and the point estimate is substantially larger than for other subcomponents. The coefficient for central bank lending remains statistically significant when we include the index for other subcomponents (column 8), and, as expected, its magnitude increases when we omit our measure of de facto lending (column 9).

B. Evidence from Local Projections

Thus far we have presented evidence of the average relationship between the CBI index and inflation. However, inflation can be persistent and institutional changes can take time to materialize, which means that the benefits of a more independent central bank in terms of reducing inflation can take time to materialize.

^{Table 5.}

Central Bank Independence and Inflation: by Subcomponent

Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1

^{Table 5.}

Central Bank Independence and Inflation: by Subcomponent

Dependent variable: inflation	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)
Central Bank Governance (CBG)	-0.0460***	-0.0297**
	(0.0110)	(0.0113)
CBI except CGB		-0.0897*** (0.0222)
Central Bank Policy Objective (CBPO)			-0.0402***	-0.00493
			(0.00942)	(0.0163)
CBI except CBPO				-0.129** (0.0531)
Central Bank Policy Formulation (CBPF)					-0.0510** -0.0259	(0.0207) (0.0208)
CBI except CBPF						-0.0918*** (0.0223)
Central Bank Lending (CBL)							-0.0933***	-0.0550**	-0.0624***
							(0.0245)	(0.0244)	(0.0188)
CBI except CBL								-0.0615**	-0.0581**
								(0.0220)	(0.0209)
Inflation (t-1)	0.569***	0.552***	0.563***	0.553***	0.566***	0.552***	0.570***	0.552***	0.569***
	(0.0467)	(0.0464)	(0.0461)	(0.0469)	(0.0481)	(0.0467)	(0.0475)	(0.0468)	(0.0484)
Change in Central Bank Claims	0.0214***	0.0238***	0.0198***	0.0248***	0.0199***	0.0235***	0.0239***	0.0236***
to Central Govt. (% of Base Money, t-1)	(0.00459)	(0.00460)	(0.00443)	(0.00499)	(0.00451)	(0.00454)	(0.00489)	(0.00458)
Output gap (t-1)	0.598***	0.611***	0.602***	0.615***	0.598***	0.611***	0.620***	0.611***	0.568***
	(0.163)	(0.163)	(0.164)	(0.163)	(0.164)	(0.163)	(0.163)	(0.162)	(0.154)
Fixed exchange rate dummy (t-1)	-12.24***	-11.98***	-12.48***	-11.86***	-12.16***	-11.97***	-12.14***	-12.00***	-11.94***
	(1.799)	(1.761)	(1.839)	(1.805)	(1.820)	(1.755)	(1.875)	(1.771)	(1.801)
Semi flexible exchange rate dummy (t-1)	-10.97***	-10.39***	-10.70***	-10.40***	-10.97***	-10.41***	-10.69***	-10.40***	-10.35***
	(1.775)	(1.813)	(1.768)	(1.844)	(1.839)	(1.833)	(1.894)	(1.819)	(1.837)
Flexible exchange rate dummy (t-1)	-9.926***	-9.430***	-9.814***	-9.361***	-9.725***	-9.411***	-9.595***	-9.432***	-9.395***
	(1.798)	(1.773)	(1.790)	(1.782)	(1.749)	(1.775)	(1.753)	(1.777)	(1.788)
Banking crisis dummy (t-1)	-0.668	0.0808	-0.256	0.168	-0.351	0.113	0.152	0.0989	0.0280
	(1.616)	(1.516)	(1.640)	(1.455)	(1.562)	(1.501)	(1.393)	(1.482)	(1.388)
US inflation (t-1)	0.334***	0.241**	0.296***	0.244**	0.349***	0.247**	0.298***	0.243**	0.201**
	(0.0958)	(0.0906)	(0.0899)	(0.0906)	(0.0841)	(0.0881)	(0.0897)	(0.0906)	(0.0907)
Constant	16.52***	22.19***	16.25***	23.47***	17.92***	22.24***	20.83***	22.03***	22.39***
	(1.968)	(3.019)	(2.130)	(3.801)	(2.819)	(3.128)	(2.876)	(2.977)	(2.762)
Country Fixed Effects	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Observations	1,133	1,133	1,133	1,133	1,133	1,133	1,133	1,133	1,161
R-squared	0.704	0.711	0.706	0.711	0.705	0.711	0.707	0.711	0.714
Number of id	17	17	17	17	17	17	17	17	17

Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1

View Table

^{Table 5.}

Central Bank Independence and Inflation: by Subcomponent

Dependent variable: inflation	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)
Central Bank Governance (CBG)	-0.0460***	-0.0297**
	(0.0110)	(0.0113)
CBI except CGB		-0.0897*** (0.0222)
Central Bank Policy Objective (CBPO)			-0.0402***	-0.00493
			(0.00942)	(0.0163)
CBI except CBPO				-0.129** (0.0531)
Central Bank Policy Formulation (CBPF)					-0.0510** -0.0259	(0.0207) (0.0208)
CBI except CBPF						-0.0918*** (0.0223)
Central Bank Lending (CBL)							-0.0933***	-0.0550**	-0.0624***
							(0.0245)	(0.0244)	(0.0188)
CBI except CBL								-0.0615**	-0.0581**
								(0.0220)	(0.0209)
Inflation (t-1)	0.569***	0.552***	0.563***	0.553***	0.566***	0.552***	0.570***	0.552***	0.569***
	(0.0467)	(0.0464)	(0.0461)	(0.0469)	(0.0481)	(0.0467)	(0.0475)	(0.0468)	(0.0484)
Change in Central Bank Claims	0.0214***	0.0238***	0.0198***	0.0248***	0.0199***	0.0235***	0.0239***	0.0236***
to Central Govt. (% of Base Money, t-1)	(0.00459)	(0.00460)	(0.00443)	(0.00499)	(0.00451)	(0.00454)	(0.00489)	(0.00458)
Output gap (t-1)	0.598***	0.611***	0.602***	0.615***	0.598***	0.611***	0.620***	0.611***	0.568***
	(0.163)	(0.163)	(0.164)	(0.163)	(0.164)	(0.163)	(0.163)	(0.162)	(0.154)
Fixed exchange rate dummy (t-1)	-12.24***	-11.98***	-12.48***	-11.86***	-12.16***	-11.97***	-12.14***	-12.00***	-11.94***
	(1.799)	(1.761)	(1.839)	(1.805)	(1.820)	(1.755)	(1.875)	(1.771)	(1.801)
Semi flexible exchange rate dummy (t-1)	-10.97***	-10.39***	-10.70***	-10.40***	-10.97***	-10.41***	-10.69***	-10.40***	-10.35***
	(1.775)	(1.813)	(1.768)	(1.844)	(1.839)	(1.833)	(1.894)	(1.819)	(1.837)
Flexible exchange rate dummy (t-1)	-9.926***	-9.430***	-9.814***	-9.361***	-9.725***	-9.411***	-9.595***	-9.432***	-9.395***
	(1.798)	(1.773)	(1.790)	(1.782)	(1.749)	(1.775)	(1.753)	(1.777)	(1.788)
Banking crisis dummy (t-1)	-0.668	0.0808	-0.256	0.168	-0.351	0.113	0.152	0.0989	0.0280
	(1.616)	(1.516)	(1.640)	(1.455)	(1.562)	(1.501)	(1.393)	(1.482)	(1.388)
US inflation (t-1)	0.334***	0.241**	0.296***	0.244**	0.349***	0.247**	0.298***	0.243**	0.201**
	(0.0958)	(0.0906)	(0.0899)	(0.0906)	(0.0841)	(0.0881)	(0.0897)	(0.0906)	(0.0907)
Constant	16.52***	22.19***	16.25***	23.47***	17.92***	22.24***	20.83***	22.03***	22.39***
	(1.968)	(3.019)	(2.130)	(3.801)	(2.819)	(3.128)	(2.876)	(2.977)	(2.762)
Country Fixed Effects	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Observations	1,133	1,133	1,133	1,133	1,133	1,133	1,133	1,133	1,161
R-squared	0.704	0.711	0.706	0.711	0.705	0.711	0.707	0.711	0.714
Number of id	17	17	17	17	17	17	17	17	17

Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1

Figure 6 shows the dynamic impact of improvements in central bank independence on inflation in Latin America. The estimated local projections suggest that a one standard deviation increase in the CBI index is associated with a reduction in inflation of approximately 1 percentage point and yields an accumulated reduction in prices of roughly 5 percentage points after 5 years.

View Full Size

The Dynamic Impact of Central Bank independence in Latin America: Baseline Results

Citation: IMF Working Papers 2022, 186; 10.5089/9798400219030.001.A001

Note: Standard errors are clustered at the country level.

• Download Figure
• Download figure as PowerPoint slide

The dynamic response of prices to changes in the CBI index does not appear to have been affected by the sweeping reforms of the 1990s. Figure 7, panel A, shows that the estimated dynamic impact of the CBI index is roughly the same in pre- and post-1990 period. Figure 7, panel B, does point to slightly higher responsiveness in the pre-1970s period, but differences between the two subperiods do not appear to be statistically significant.

The initial response of inflation to changes in the CBI appears to be stronger when other areas of reform are less advanced, but differences disappear after a few years. As shown in Figure 8, inflation declines gradually when CBI is improved in countries where the structural reforms index takes low values, with a cumulative reduction in prices of close to 5 percent after 5 years. In countries with high levels of the structural reforms index, prices remain flat initially, but eventually decline to levels comparable to those observed in countries with low levels of the structural reform index.

Next, we study the extent to which results are driven by the initial level of inflation. This may be an important consideration if changes to central bank laws are introduced in response to inflationary episodes.

View Full Size

The Dynamic Impact of Central Bank independence in Latin America: Differences Over Time

Citation: IMF Working Papers 2022, 186; 10.5089/9798400219030.001.A001

Note: Standard errors are clustered at the country level.

• Download Figure
• Download figure as PowerPoint slide

Figure 9 shows that, while inflation responds strongly to improvements in CBI when inflation takes above average levels, CBI yields cumulative declines in prices even in low-inflation environments. The estimated coefficients suggest a 20 percentage point cumulative decline in prices after 5 years following a one standard deviation improvement in the CBI when inflation is initially high (Panel A). When inflation is low, the decline is more modest, but it is statistically significant (Panel B). The fact that prices respond strongly when inflation is high is not surprising—after all, these are typically situations where inflation inertia is high, and changes in CBI work as shock therapy to break the pattern. What is perhaps more surprising is the fact that CBI’s bite is also present in low inflation settings when arguably expectations are better anchored.

Taken together, our results point to a strong dynamic link between central bank independence and prices in Latin America. The estimated impacts suggest that CBI remains an important policy lever to achieve price stability, regardless of the state of the economy or the advancement of other policy reform areas. They also highlight the potential adverse effects of an erosion of independence in terms of inflation. Importantly, Latin America’s history suggests that a deterioration of central bank independence can result not only in a gradual rise in inflation but could increase the likelihood of inflation becoming unanchored and of a high inflation episode materializing, a concern that we tackle empirically next.

View Full Size

The Dynamic Impact of Central Bank independence in Latin America–by Structural Reform Level

Citation: IMF Working Papers 2022, 186; 10.5089/9798400219030.001.A001

Note: Standard errors are clustered at the country level.

• Download Figure
• Download figure as PowerPoint slide

C. CBI and inflation tail risks

The results from the standard panel regressions and from local projections assess the link between CBI levels and average inflation. However, one potential benefit of an independent central bank focused on combating inflation is that it can prevent extreme episodes of inflation. The individual experience of Latin American countries showcased in previous sections provides suggestive evidence to this potential benefit of central bank independence, as arguably the loosening of CBI was a harbinger of future episodes on high inflation.

To study whether this narrative evidence is systematically supported in the data, we pursue two separate empirical exercises. The first exercise estimates quantile regressions akin to those in equation (1). This approach aims at estimating the impact of CBI not only on average inflation but on different moments of the inflation distribution. If the narrative evidence presented earlier is robustly seen in the data, one would expect that CBI has a larger coefficient at the right tail of the inflation distribution.

View Full Size

The Dynamic Impact of Central Bank independence in Latin America—by initial levels of inflation

Citation: IMF Working Papers 2022, 186; 10.5089/9798400219030.001.A001

Note: Standard errors are clustered at the country level.

• Download Figure
• Download figure as PowerPoint slide

Figure 10, Panel A provides some support to the role of CBI in reducing inflationary tail risks. Indeed, our quantile regressions (estimated at each decile of the inflation distribution) points to a relative stable and modest, albeit statically significant, link between CBI and inflation up to the 5th decile of the inflation distribution in the sense that the coefficient is relatively small. By contrast, coefficients for deciles above the median are noticeably larger in absolute value for larger deciles, with the CBI coefficient for the 90th percentile being almost 2 times as large as that for the median. The results for the overall index resemble those for the subcomponent on central bank lending to the central government (Panel B), while the coefficient from the quantile regression exercise for other subcomponents appear to be much more stable across quantiles.

View Full Size

CBI’s Impact Along the Inflation Distribution

Citation: IMF Working Papers 2022, 186; 10.5089/9798400219030.001.A001

• Download Figure
• Download figure as PowerPoint slide

Similar conclusions emerge from the estimation of a linear probability model. In particular, we define episodes of high inflation as years when inflation crosses a specific threshold. For robustness purposes, we define three high inflation thresholds, the 80th percentile of inflation in our sample, the 90th percentile, and the 95th percentile. The results from these exercises, which are presented in Table 6, show that higher values of the CBI index reduce the likelihood of high inflation episodes, and in all cases the estimated coefficient is statistically significant. The result suggests that a 30-point increase in the CBI, roughly equivalent to moving from the 25th to the 75th percentile of the CBI distribution, results, all else equal, in reductions in the likelihood of crossing the high inflation threshold that range from 8 percentage points (in the case of the 80th percentile threshold) to close to 5 percentage points (in the case of the 95th percentile threshold). The dimmed importance of CBI in reducing the likelihood of high inflation outcomes as we increase the threshold appears to be related to the increased preponderance of central bank lending to the central government for higher values of the inflation threshold. This stresses the crucial role played by de facto limits to central bank lending and the perils of fiscal dominance.

An exploration of the role of each subcomponent in preventing high inflation episodes confirms the importance of de facto limits to central bank lending (Table 7). All subcomponents of the index are negatively associated with episodes of high inflation at low thresholds, suggesting that higher values of each of the subcomponents reduces the likelihood of these episodes. At the highest high-inflation threshold, however, the point estimates for all subcomponents drop substantially, in some cases losing statistical significance, and the importance of de facto lending increases.

^{Table 6.}

Central Bank Independence and Inflation

Standard errors clustered at the country level *** p<0.01, ** p<0.05, * p<0.1

^{Table 6.}

Central Bank Independence and Inflation

Dependent variable: High inflation dummy	(1) 80th pctile threshold	(2) 90th pctile threshold	(3) 95th pctile threshold
CBI	-0.00287** (0.00125)	-0.00177* (0.000988)	-0.00149* (0.000894)
Inflation (t-1)	8.18e-05*** (2.18e-05)	5.27e-05** (2.43e-05)	3.22e-05 (3.73e-05)
Change in Central Bank Claims to Central Govt. (% of Base Money, t-1)	1.84e-05 (0.000119)	0.000269** (0.000132)	0.000866*** (0.000183)
Output gap (t-1)	0.0161*** (0.00448)	0.0122*** (0.00345)	0.00339 (0.00326)
Banking crisis dummy (t-1)	0.0871** (0.0391)	0.0951** (0.0417)	0.0457 (0.0322)
Fixed exchange rate dummy (t-1)	-0.647*** (0.0383)	-0.597*** (0.0395)	-0.302*** (0.0401)
Semi flexible exchange rate (t-1)	-0.624*** (0.0436)	-0.582*** (0.0382)	-0.288*** (0.0388)
Flexible exchange rate (t-1)	-0.596*** (0.0445)	-0.592*** (0.0441)	-0.302*** (0.0410)
US inflation (t-1)	0.00726* (0.00386)	0.00298 (0.00237)	0.00371 (0.00323)
Constant	0.848*** (0.0939)	0.713*** (0.0884)	0.392*** (0.0846)
Observations	1,133	1,133	1,133
Number of id	17	17	17

Standard errors clustered at the country level *** p<0.01, ** p<0.05, * p<0.1

View Table

^{Table 6.}

Central Bank Independence and Inflation

Dependent variable: High inflation dummy	(1) 80th pctile threshold	(2) 90th pctile threshold	(3) 95th pctile threshold
CBI	-0.00287** (0.00125)	-0.00177* (0.000988)	-0.00149* (0.000894)
Inflation (t-1)	8.18e-05*** (2.18e-05)	5.27e-05** (2.43e-05)	3.22e-05 (3.73e-05)
Change in Central Bank Claims to Central Govt. (% of Base Money, t-1)	1.84e-05 (0.000119)	0.000269** (0.000132)	0.000866*** (0.000183)
Output gap (t-1)	0.0161*** (0.00448)	0.0122*** (0.00345)	0.00339 (0.00326)
Banking crisis dummy (t-1)	0.0871** (0.0391)	0.0951** (0.0417)	0.0457 (0.0322)
Fixed exchange rate dummy (t-1)	-0.647*** (0.0383)	-0.597*** (0.0395)	-0.302*** (0.0401)
Semi flexible exchange rate (t-1)	-0.624*** (0.0436)	-0.582*** (0.0382)	-0.288*** (0.0388)
Flexible exchange rate (t-1)	-0.596*** (0.0445)	-0.592*** (0.0441)	-0.302*** (0.0410)
US inflation (t-1)	0.00726* (0.00386)	0.00298 (0.00237)	0.00371 (0.00323)
Constant	0.848*** (0.0939)	0.713*** (0.0884)	0.392*** (0.0846)
Observations	1,133	1,133	1,133
Number of id	17	17	17

Standard errors clustered at the country level *** p<0.01, ** p<0.05, * p<0.1

^{Table 7.}

Central Bank Independence and High Inflation Episodes, by CBI sub-components

Standard errors clustered at the country level in parentheses *** p<0.01, ** p<0.05, * p<0.1

^{Table 7.}

Central Bank Independence and High Inflation Episodes, by CBI sub-components

Dep. Variable: High inflation dummy	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)	(11)	(12)
	50 percent threshold				100 percent threshold				150 perent threshold
Central Bank Governance	-0.000747				0.000103				-0.000763
	(0.000637)				(0.000548)				(0.000492)
Central Bank Policy Objective		-0.000918**				-0.000349				-0.000271*
		(0.000439)				(0.000248)				(0.000151)
Central Bank Policy Formulation			-0.00173**				-0.00112				-0.000833
			(0.000867)				(0.000856)				(0.000602)
Central Bank Lending				-0.00186*				-0.00148**				-0.00100*
				(0.00105)				(0.000745)				(0.000603)
Inflation (t-1)	8.29e-05***	8.15e-05***	8.60e-05***	8.25e-05***	5.36e-05**	5.31e-05**	5.55e-05**	5.26e-05**	3.23e-05	3.24e-05	3.42e-05	3.27e-05
	(2.18e-05)	(2.34e-05)	(2.23e-05)	(2.20e-05)	(2.40e-05)	(2.38e-05)	(2.41e-05)	(2.47e-05)	(3.74e-05)	(3.71e-05)	(3.75e-05)	(3.71e-05)
Net Central Bank Claims on	-2.84e-05	-5.54e-05	-0.000110	3.86e-05	0.000227	0.000229	0.000187	0.000298**	0.000848***	0.000826***	0.000804***	0.000881***
Central Government (% of Money, t-1)	(0.000123)	(0.000131)	(0.000130)	(0.000132)	(0.000140)	(0.000139)	(0.000148)	(0.000130)	(0.000187)	(0.000193)	(0.000197)	(0.000182)
Output gap (t-1)	0.0156***	0.0158***	0.0155***	0.0160***	0.0119***	0.0120***	0.0118***	0.0123***	0.00320	0.00316	0.00310	0.00339
	(0.00444)	(0.00436)	(0.00438)	(0.00447)	(0.00348)	(0.00340)	(0.00347)	(0.00339)	(0.00325)	(0.00320)	(0.00318)	(0.00324)
Banking Crisis Dummy (t-1)	0.0708*	0.0790*	0.0763*	0.0875**	0.0889**	0.0893**	0.0874**	0.0991**	0.0371	0.0404	0.0397	0.0464
	(0.0393)	(0.0409)	(0.0390)	(0.0369)	(0.0394)	(0.0409)	(0.0401)	(0.0397)	(0.0313)	(0.0317)	(0.0307)	(0.0313)
Fixed Exchange Rate Dummy (t-1)	-0.667***	-0.665***	-0.641***	-0.662***	-0.619***	-0.611***	-0.590***	-0.605***	-0.308***	-0.314***	-0.300***	-0.310***
	(0.0437)	(0.0387)	(0.0384)	(0.0364)	(0.0472)	(0.0431)	(0.0415)	(0.0396)	(0.0418)	(0.0468)	(0.0409)	(0.0444)
Semi flexible Exchange Rate Dummy (t-1)	-0.652***	-0.637***	-0.625***	-0.643***	-0.610***	-0.597***	-0.579***	-0.593***	-0.300***	-0.300***	-0.290***	-0.298***
	(0.0412)	(0.0393)	(0.0462)	(0.0362)	(0.0423)	(0.0387)	(0.0422)	(0.0369)	(0.0403)	(0.0441)	(0.0397)	(0.0423)
Flexible Exchange Rate Dummy (t-1)	-0.616***	-0.606***	-0.600***	-0.613***	-0.614***	-0.603***	-0.592***	-0.603***	-0.310***	-0.313***	-0.304***	-0.311***
	(0.0509)	(0.0462)	(0.0446)	(0.0409)	(0.0514)	(0.0476)	(0.0452)	(0.0435)	(0.0439)	(0.0468)	(0.0420)	(0.0440)
US Inflation (t-1)	0.0102**	0.00876**	0.00924**	0.00917**	0.00572**	0.00455*	0.00417	0.00375	0.00464	0.00509	0.00484	0.00466
	(0.00435)	(0.00391)	(0.00399)	(0.00386)	(0.00277)	(0.00267)	(0.00264)	(0.00275)	(0.00319)	(0.00373)	(0.00327)	(0.00374)
Constant	0.703***	0.708***	0.783***	0.799***	0.597***	0.617***	0.674***	0.708***	0.337***	0.310***	0.353***	0.370***
	(0.0456)	(0.0499)	(0.0727)	(0.0920)	(0.0412)	(0.0523)	(0.0761)	(0.0852)	(0.0618)	(0.0446)	(0.0668)	(0.0716)
Country Fixed Effects	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Observations	1,133	1,133	1,133	1,133	1,133	1,133	1,133	1,133	1,133	1,133	1,133	1,133
Number of id	17	17	17	17	17	17	17	17	17	17	17	17

Standard errors clustered at the country level in parentheses *** p<0.01, ** p<0.05, * p<0.1

View Table

^{Table 7.}

Central Bank Independence and High Inflation Episodes, by CBI sub-components

Dep. Variable: High inflation dummy	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)	(11)	(12)
	50 percent threshold				100 percent threshold				150 perent threshold
Central Bank Governance	-0.000747				0.000103				-0.000763
	(0.000637)				(0.000548)				(0.000492)
Central Bank Policy Objective		-0.000918**				-0.000349				-0.000271*
		(0.000439)				(0.000248)				(0.000151)
Central Bank Policy Formulation			-0.00173**				-0.00112				-0.000833
			(0.000867)				(0.000856)				(0.000602)
Central Bank Lending				-0.00186*				-0.00148**				-0.00100*
				(0.00105)				(0.000745)				(0.000603)
Inflation (t-1)	8.29e-05***	8.15e-05***	8.60e-05***	8.25e-05***	5.36e-05**	5.31e-05**	5.55e-05**	5.26e-05**	3.23e-05	3.24e-05	3.42e-05	3.27e-05
	(2.18e-05)	(2.34e-05)	(2.23e-05)	(2.20e-05)	(2.40e-05)	(2.38e-05)	(2.41e-05)	(2.47e-05)	(3.74e-05)	(3.71e-05)	(3.75e-05)	(3.71e-05)
Net Central Bank Claims on	-2.84e-05	-5.54e-05	-0.000110	3.86e-05	0.000227	0.000229	0.000187	0.000298**	0.000848***	0.000826***	0.000804***	0.000881***
Central Government (% of Money, t-1)	(0.000123)	(0.000131)	(0.000130)	(0.000132)	(0.000140)	(0.000139)	(0.000148)	(0.000130)	(0.000187)	(0.000193)	(0.000197)	(0.000182)
Output gap (t-1)	0.0156***	0.0158***	0.0155***	0.0160***	0.0119***	0.0120***	0.0118***	0.0123***	0.00320	0.00316	0.00310	0.00339
	(0.00444)	(0.00436)	(0.00438)	(0.00447)	(0.00348)	(0.00340)	(0.00347)	(0.00339)	(0.00325)	(0.00320)	(0.00318)	(0.00324)
Banking Crisis Dummy (t-1)	0.0708*	0.0790*	0.0763*	0.0875**	0.0889**	0.0893**	0.0874**	0.0991**	0.0371	0.0404	0.0397	0.0464
	(0.0393)	(0.0409)	(0.0390)	(0.0369)	(0.0394)	(0.0409)	(0.0401)	(0.0397)	(0.0313)	(0.0317)	(0.0307)	(0.0313)
Fixed Exchange Rate Dummy (t-1)	-0.667***	-0.665***	-0.641***	-0.662***	-0.619***	-0.611***	-0.590***	-0.605***	-0.308***	-0.314***	-0.300***	-0.310***
	(0.0437)	(0.0387)	(0.0384)	(0.0364)	(0.0472)	(0.0431)	(0.0415)	(0.0396)	(0.0418)	(0.0468)	(0.0409)	(0.0444)
Semi flexible Exchange Rate Dummy (t-1)	-0.652***	-0.637***	-0.625***	-0.643***	-0.610***	-0.597***	-0.579***	-0.593***	-0.300***	-0.300***	-0.290***	-0.298***
	(0.0412)	(0.0393)	(0.0462)	(0.0362)	(0.0423)	(0.0387)	(0.0422)	(0.0369)	(0.0403)	(0.0441)	(0.0397)	(0.0423)
Flexible Exchange Rate Dummy (t-1)	-0.616***	-0.606***	-0.600***	-0.613***	-0.614***	-0.603***	-0.592***	-0.603***	-0.310***	-0.313***	-0.304***	-0.311***
	(0.0509)	(0.0462)	(0.0446)	(0.0409)	(0.0514)	(0.0476)	(0.0452)	(0.0435)	(0.0439)	(0.0468)	(0.0420)	(0.0440)
US Inflation (t-1)	0.0102**	0.00876**	0.00924**	0.00917**	0.00572**	0.00455*	0.00417	0.00375	0.00464	0.00509	0.00484	0.00466
	(0.00435)	(0.00391)	(0.00399)	(0.00386)	(0.00277)	(0.00267)	(0.00264)	(0.00275)	(0.00319)	(0.00373)	(0.00327)	(0.00374)
Constant	0.703***	0.708***	0.783***	0.799***	0.597***	0.617***	0.674***	0.708***	0.337***	0.310***	0.353***	0.370***
	(0.0456)	(0.0499)	(0.0727)	(0.0920)	(0.0412)	(0.0523)	(0.0761)	(0.0852)	(0.0618)	(0.0446)	(0.0668)	(0.0716)
Country Fixed Effects	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Observations	1,133	1,133	1,133	1,133	1,133	1,133	1,133	1,133	1,133	1,133	1,133	1,133
Number of id	17	17	17	17	17	17	17	17	17	17	17	17

Standard errors clustered at the country level in parentheses *** p<0.01, ** p<0.05, * p<0.1