Central Bank Governance and Reserve Portfolios Investment Policies: An Empirical Analysis

This paper uses a unique survey data set of 105 central banks to investigate whether investment policies for central bank foreign reserve portfolios are linked to the governance arrangements for reserve management. The paper evaluates whether a central bank's investment decision-making structure impacts how much risk institutions take in their reserve management operations and the level of diversity in their reserve portfolios. Additionally, it explores the implications of the broader governance environment on reserve management. The analysis yields four key findings. First, internal governance arrangements matter for foreign reserve portfolio investment policy; the empirical results indicate that reserve portfolios are more diversified in central banks in which the middle office directly reports to the board. Second, controlling for the level of reserves, the macroenvironment, and the broader governance environment, reserve portfolios are more diversified in central banks where the back, middle, and front offices are separated. Third, the regression analysis also reveals that central banks in countries where the Ministry of Finance has an obligation to cover negative equity have fewer eligible currencies and are therefore less diversified. Fourth, central banks where boards actively exercise portfolio oversight usually have portfolios with more risk and diversification. Portfolios with longer investment horizons, more currencies, and a broader set of asset classes have performed better historically while limiting downside risk. Given that the analysis controls the broader governance environment, the data indicate that any central bank can improve its internal governance regardless of the external governance environment. This paper contributes to the literature on central bank foreign reserves management and on understanding the importance of governance arrangements in investment policy.


Introduction
Governance is crucial for central banks. According to Bossu and Rossi (2019), the concept of central bank governance is multifaceted. They contend that a clearly defined central bank mandate and decision-making structure are essential to support accountability and legitimacy. Such a mandate should elaborate on objectives, functions, and powers, specifying the legal tools available for a central bank to implement its functions. Clear decision-making structures are also necessary to ensure adequate implementation of the mandate. Given the complexity of central banks and the multiple responsibilities they assume, it is crucial to define what to decide, who must decide, and how decisions must be made (Bossu and Rossi 2019).
Our empirical analysis focuses on the governance of reserve management, a key part of central bank governance overall. As foreign currency reserves have significantly increased and represent an essential part of central bank assets, the process for making investment decisions and overseeing results becomes more relevant. We aim to contribute to the reserve management governance discussion with data-driven analysis using a unique survey data set on central bank reserve management practices. Most previous publications on the topic are prescriptive and qualitative. Notably, our research shows that governance arrangements do matter for the investment policy of foreign reserve portfolios.
As the first step in this analysis, we briefly review the literature on central bank governance, focusing on reserve management operations. We then describe the data that we use for empirical research. We use a components of their investment policies and asset allocations, and their accounting and profit-sharing methodologies. We also deploy data that describes the broader institutional and macroenvironment in which these central banks operate and proxies for reserve adequacies. We use this data set to address novel research questions to find links between specific governance arrangements and reserve management investment policies. Finally, we conclude and discuss the possible implications of our key findings for the management of reserves.
Electronic copy available at: https://ssrn.com/abstract=3914415 backed by gold, and their reserves primarily consisted of gold bullion. With the abolition of the gold standard in 1971, central banks started diversifying their reserves into foreign asset pools.
Holding foreign reserves has various purposes. Most notable are self-insurance for balance-of-payments crises and support of exchange rate policies. In some countries, reserves also support financial stability by providing lender-of-last-resort functions in foreign currency. As a result, reserves protect the country against disruptions and volatility in international capital markets. Over the past two decades, successive financial crises have solidified and validated this approach. Recent research indicates that emerging market central banks holding relatively higher foreign reserves experienced less currency depreciation, smoother credit growth, and more stable credit ratings. Countries with higher levels of reserves also had better access to external funding during the global financial crisis of 2008-09 (Arslan and Cantú 2019). Furthermore, empirical research identifies a relationship between political instability and weak institutions in emerging markets, resulting in central banks accumulating international reserves as a risk mitigation method to deal with broader economic uncertainties (de la Rosa, 2015).
Foreign reserves have grown significantly over the past four decades, and they are now at record highs ( Figure 1). Emerging economies, notably China, have led this trend. Economic development and higher commodity prices have encouraged capital inflows and higher exports in emerging economies. As a result, central banks have intervened in foreign exchange markets and accumulated foreign reserves to curb the appreciation of their exchange rates. As Figure 1 shows, this build-up has been a worldwide phenomenon.
On average, emerging market central banks 5 have accumulated foreign reserve levels at or above 23 percent of GDP (see Figure 1). 6 Many countries hold reserves far above traditional benchmarks, such as three months of import cover, 20 percent of broad money, and 100 percent cover of short-term external debt repayments. In addition, according to the Assessing Reserve Adequacy (ARA) metric 7 of the International Monetary Fund (IMF), approximately half of the reported countries have reserves above the adequate level.

Figure 1 Accumulation of Foreign Reserves Worldwide
The buildup of reserves has transformed the way central banks manage these funds. Although they remain invested primarily in government bonds and other conservative instruments, the adoption of nontraditional asset classes, such as mortgage-backed securities, corporate bonds, and equities, is on the rise (Anasashvili et al. 2020 Furthermore, the role of every person who participates in the investment process must be defined clearly, along with well-defined documentation of processes, to allow continuous decision-making (Ruiz 2020).
RAMP (Anasashvili et al. 2020) collected data on how central banks organize their reserve management operations. The survey findings indicate that reserve governance arrangements vary across central banks.
Most of the 105 respondents to the 2019 RAMP survey follow a three-tier governance structure. Ninetytwo percent of central banks reported that their respective boards approved the reserve management policy, including high-level decisions such as reserve management objectives, risk tolerance, investment horizon, and strategic asset allocation. Moreover, RAMP (Anasashvili et al. 2020) also showed that central banks' boards frequently approve the investment management guidelines, that is, the specific investment rules for managing the portfolio, indicating that many central bank investment committees have limited decisionmaking power. Many boards also have the responsibility of hiring external managers (Anasashvili et al. 2020). Finally, the survey showed that middle office reporting to the board on risk and return information varies across central banks (see Anasashvili et al. 2020).
Apart from the specific governance arrangements for reserves management operations, the RAMP survey

Approach and Objectives
Little quantitative research is available on the link between overall central bank governance and reserve management, despite the critical role that sound reserve management plays in helping, supporting, and maintaining confidence in monetary management (Al Hassan, Farahmand, and Papaianou 2014). As shown above, most publications on reserve management governance are prescriptive and qualitative. Therefore, we contribute to the reserve management governance discussion with data-driven analysis using RAMP's unique survey data on governance and organizational arrangements of central banks and asset allocation and risk measures. Specifically, we investigate whether governance arrangements impact investment policies and central bank risk taking and, if so, precisely which arrangements matter. We also analyze whether organizational arrangements impact central banks' investment policies and whether reporting Electronic copy available at: https://ssrn.com/abstract=3914415 structures influence central banks' investment policies and risk taking in their reserve management operations. This paper's ultimate goal is to empirically analyze the relationships between a central bank's governance structure for its reserve management operations and its investment policies and risk taking.

Methodology and Data
We use correlations and regression analysis to find links between specific governance arrangements and variables related to central banks' reserves investment policies. We start by analyzing correlations between the variables described above and testing for their statistical significance. We then use regression analysis to analyze whether some of the correlation results hold when adjusting for reserve adequacy and indicators that describe the macroenvironment.
We use data groups that capture the governance arrangement, investment policies, and measures of risk taking of individual central banks to test whether governance affects investment policies. First, we use data on governance arrangements for central banks' reserve management operations as an independent variable collected through the RAMP surveys. Second, we utilize RAMP survey data on the composition and risk of reserve portfolios as dependent variables describing a central bank's investment policy. We then deploy three types of control variables to isolate empirically the impact of the governance structure and investment policies.

Governance and Macroeconomic Variables
We compiled governance variables from multiple sources, collected at the national level, to assess the broader governance environment as a control variable to isolate the effect of the governance arrangements at the central bank level. We also use macroeconomic variables and data on reserve adequacy to further isolate the effects (see Table 1 for a summary).
Electronic copy available at: https://ssrn.com/abstract=3914415 This index measures the "perceptions of the quality of public services, the quality of the civil service and the degree of its independence from political pressures, the quality of policy formulation and implementation, and the credibility of the government's commitment to such policies." 8 The scale runs from -2.5 (lowest) to 2.5 (highest) and covers 204 countries. The index is developed by compiling and summarizing information from over 30 existing data sources that report the views and experiences of citizens, entrepreneurs, and experts in the public, private, and nongovernmental sectors. The rationale for using this variable is that countries with strong governance and effective public sectors usually have well-  collected by various institutions. The index has a scale of 0-100, where zero represents the highest level of perceived corruption and 100 the lowest level of perceived corruption. 10 Finally, we use the Governance Pillar Score (MSCI). This index covers 198 countries to "assess the extent to which a country's long-term competitiveness is affected by its institutional capacity to support long-term stability and functioning of its financial, judicial, and political systems, and capacity to address the environmental and social risks." 11 The Governance Pillar Score is included in the MSCI assessment on the environment, social, and governance (ESG) risk of sovereign countries. The best score is zero; the worst is ten. The index is derived from thirdparty sources and focuses on political rights and civil liberties, stability and peace, control of corruption, and public financial management.
Macroeconomic Variables. We then use the country risk score from the Economist Intelligence Unit's Country Risk Model to control for the macroenvironment. This score rates country risk on a scale from zero (no risk) to 100 (maximum risk), by taking a simple average of the sovereign, currency, and banking sector risk scores of a given country. 12 We included this variable in the analysis to assess the macroenvironment in which central banks operate. Alternatively, we use the number of currency crises and the current account balance as a percentage of GDP as another means of describing the macroenvironment in which the central bank operates.

Reserve Adequacy Variables.
Finally, a central bank's risk tolerance and investment policy (number of eligible assets and currencies, portfolio duration, allocation to nontraditional asset classes) are linked to reserve adequacy. All else being equal, a central bank with higher reserve adequacy typically has a higher risk tolerance and can deploy a broader range of asset classes, currencies, and longer duration. Therefore, we use reserve adequacy measures to control for differences in reserve adequacy across central banks. We include two reserve adequacy metrics as control variables: reserves to GDP and to M2. Separately, we also control for GDP per capita and other macroeconomic variables, such as short-term external debt stocks as a percentage of total reserves. were collected systematically across central banks globally. In addition to questions on governance arrangements and investment policies, the surveys also included questions on asset allocation, portfolio management, risk management, performance and risk reporting, and transparency. Table 2 lists the survey variables we used to test our set of questions empirically.

Composition of reserve portfolio -Continuous variables
(1) We consider an investment committee independent if it approves the investment management guidelines, that is, the specific investment rules for managing the portfolio.
(2) We explore whether the middle office reports directly to the board.
(3) We also consider whether the middle office reports performance and risk metrics to the investment committee.
(4) We test whether a central bank organizes its reserve management operations in one department.
(5) We further assess whether the reserve management operations are arranged in separate departments.
(6) Finally, we probe the impact of the Ministry of Finance's obligation to cover the central bank's negative equity.
Relatedly, we further probe transparency on the risks and returns of reserve management operations to the board as a continuous variable.
We use RAMP survey data collected as continuous data on the number of eligible currencies and asset classes and the allocation to nontraditional asset classes as dependent variables as a reflection of a central bank's investment policies. We also deploy RAMP data to compute the measures of risk as reflected in a central bank's current asset allocation: (1) estimated risk of the portfolio, (2) investment horizon (measured at the total portfolio level for both tranched and untranched portfolios), and (3) duration (measured at the total portfolio level and at the liquidity or investment tranche level).

Descriptive Statistical Results
We run simple correlations for the various independent governance variables to identify patterns in the data.
The correlations are run to develop context and show empirical relationships between critical elements of reserve management. The full results of the correlation analysis are found in Appendix I.

Independent Investment Committees.
We test whether there is a positive statistical correlation between the independence of the investment committee and the characteristics of the investment guidelines. As explained above, we define an investment committee as independent if it is responsible for approving the investment guidelines. Table 3 indicates that an independent investment committee is negatively correlated with the number of eligible currencies and asset classes. Similarly, duration is negatively correlated with an independent investment committee. Middle Office Reporting Directly to the Board. Additionally, we find a statistically positive correlation between the middle office's reporting and a central bank's risk-taking and investment policies (Table 4).
When the middle office reports directly to the board, we observe a higher allocation to nontraditional assets and more diversified reserve portfolios as the number of currencies and asset classes increases (see Figure   3). Organization of Operational Units. We do not find any statistically significant correlation between the organization of the operational units and the diversification of reserve management portfolios or their risk taking. As discussed earlier, the RAMP surveys show that central banks typically clearly separate front-, middle-, and back-office functions. However, we did not find a link between putting all functions in the same or different departments and the level of risk and diversification.

Obligation to Cover Negative Equity and Reserve Portfolio Metrics.
Our empirical analysis shows that central banks that do not receive financial support from the government tend to have a more diversified strategic asset allocation with lower overall portfolio risk, even when controlling for distribution policies (see Table 5). Countries in which finance ministries have an obligation to cover central banks' negative equity also have riskier macroenvironments. This observation is consistent with the result that the obligation to cover negative equity is more common in countries with weaker governance. It also may indicate that the necessity of government support for the central bank may be higher in such environments as the economy is less stable and prone to unforeseen shocks that the central bank is not well prepared to manage with existing resources.

Table 5 Obligation to Cover Negative Equity and Investment Policies
Note: The significance level of the pairwise correlations is displayed with asterisks, where *** stands for a p-value below 0.01 (i.e., extremely significant), ** stands for a p-value below 0.05 (i.e., very significant), and * stands for a p-value below 0.1 (i.e., significant).

The Broader Institutional Environments and Reserve Portfolio Metrics.
Our analysis finds that the broader governance environment in which central banks operate correlates with certain investment policy types (see Table 6). As the table indicates, the government effectiveness index has a strong, statistically significant positive correlation with duration risk and portfolio diversification, measured as the number of eligible asset classes and the allocation to nontraditional asset classes (see Figure 4). Because of these findings, we will use indices that describe the broader institutional environment as control variables in our regressions.

Multivariate Regression Results
We use regression analysis to delve into some of the correlation analysis on central bank governance and risk. We test the extent to which governance arrangements matter for risk taking and diversification in foreign reserves portfolios while controlling for the broader governance environment in which central banks operate, the level of reserve adequacy, the macroenvironment, or country risk. 15 Additional regression results are found in Appendix II.
15. This choice of variables was influenced by a general-to-specific regression (GETS) run as a way of selecting the most relevant variables out of a relatively large sample of variables when fitting a regression model.

Independent Investment Committee.
We examine if approving the investment guidelines at the board or the investment committee level makes any difference for reserve management operations. In most central banks, either can have this role. The most significant advantage of leaving this decision to the board is that it has more authority. However, central bank board members may not be financial experts and have less time to focus on reserve management policy. By contrast, the investment committee has less authority, but it can meet more often, and the members usually understand operational nuances and have financial expertise. We test whether the independence of an investment committee influences central bank investment policies, as suggested in our correlation exercise. As explained, as reserve adequacy, the general macroenvironment, and government effectiveness may influence a central bank's risk taking as expressed in their investment policy, we control for these factors. We do not find with any statistical significance that having the investment committee approve guidelines matters, even when different control variables are used to describe the macroenvironment (see Table 7). considering that we control for reserve adequacy and a country's macroenvironment as well as the broader governance environment (see Table 8). These results are robust when using different specifications for macro risks (see Appendix II), and the results also hold when subsampling the data by income level and level of reserves. 17 However, there does not seem to be a significant difference between the groups.

Organization of Reserve Management Operations and Investment Policy.
We find support that how institutions organize their reserve management operations matters for the investment policy of central banks. Central banks in which the front, middle, and back office are in the same department have, on average, significantly shorter investment horizons. As macrovolatility, levels of reserves, and the broader governance environment may influence the investment horizon of central banks' reserve operation, we include these in the regression as control variables (see Table 9).
17. The data was divided into high-income countries and low-income countries, on the one hand, and by level of reserves, above and below US$15 billion, on the other hand. 18 Each of the columns in this table refers to a different regression with the column name as dependent variable (number of eligible asset classes, number of eligible currencies, estimated risk of the portfolio, and investment horizon).

Obligation of the Ministry of Finance to Cover Central Bank Negative Equity. The distribution of profits
and losses within the government is a sensitive issue for central banks. Although central banks commonly share their earnings with ministries of finance, there are questions regarding the source of those profits (i.e., foreign exchange revaluation or changes in market prices) and the timing (i.e., realized versus unrealized).
Despite significant variation in their practices, central banks tend to make partial distributions of realized profits. Our regression analysis confirms that the finance ministry's obligation to cover central banks' negative equity influences investment policy. Reserve portfolios in countries where finance ministries must cover negative equity are less diversified in terms of eligible currencies, even when controlling for reserve adequacy and the macro and broader governance environment. On average, central banks operating in countries where the Ministry of Finance had an obligation to cover negative equity held 2.4 to 2.7 fewer eligible currencies across the various regression specifications (see Table 10). It appears that this obligation impacts investment policies narrowly, as we cannot find any robust results for asset diversification or level of risks, nor for the investment horizon.

Conclusions and Policy Implications
Effective governance is essential for central banks and their reserve management function. Multiple authors have analyzed central bank governance arrangements around the world. These publications conclude that governance practices vary from country to country, but some principles seem broadly relevant for achieving positive outcomes. Notably, empirical evidence points to the importance of central bank independence and transparency for monetary policy. Different publications agree that setting up a proper governance structure is the cornerstone of a successful investment operation in reserve management. However, most publications on reserve management governance are prescriptive and qualitative.
We contribute to the discussion on reserve management governance with data-driven analysis. Our novel empirical study indicates that specific governance arrangements impact investment policy and risk taking in central banks' reserve management operations, controlling for the macroeconomic environment, reserve levels, and the broader governance environment.
Electronic copy available at: https://ssrn.com/abstract=3914415 We find that three types of governance factors relate to diversification and risk in foreign reserve portfolios.
First, direct communication between the board and the middle office often coincides with more diversified reserve operations. Notably, the result holds when controlling for reserve adequacy, and country risk indicates that anchoring risk taking at the board level allows reserve managers to have more diversified portfolios in terms of eligible assets and eligible currencies, everything else equal. Second, we find that the organizational structure for reserve operations has an impact on investment policy. Having all three units responsible for managing the reserve operations in the same department, with the same reporting line, affects central banks' ability to take risks, as reflected in the investment horizon. Controlling for the macro and governance environment and reserve adequacy, central banks with the same reporting lines for the front, middle, and back office, on average, have a significantly shorter investment horizon. We also find that this result is robust across different specifications. Third, countries where the Ministry of Finance is obligated to cover negative equity have, on average, investment policies with fewer eligible currencies than do countries where the Ministry of Finance does not have such obligations.
The most important policy implication of our analysis is the critical role of the board in reserve management. Central banks where boards actively exercise portfolio oversight usually have portfolios with more risk and diversification. While the ability and tolerance for risk taking vary across central banks, portfolios with longer investment horizons, more currencies, and more asset classes have performed better historically, while keeping downside risk limited. Given that we control the broader governance environment, our data indicate that any central bank can improve its internal governance regardless of the external governance environment. Several central banks in our database have implemented robust reserve management practices, even without optimal external governance environments.

Financial Supervision and the Broader Governance Environment
Note: The significance level of the pairwise correlations is displayed with asterisks, where *** stands for a p-value below 0.01 (i.e., extremely significant), ** stands for a p-value below 0.05 (i.e., very significant), and * stands for a p-value below 0.1 (i.e., significant).
Regarding the impact on reserve management, the correlations suggest that countries where the central bank is also a financial supervisor tend to have an independent investment committee to which the middle office reports (see Table III.2). This arrangement may result in a lower risk appetite in managing reserves and imply that the central bank accounts for and manages the risks of a more volatile governance environment in which it operates. As other findings of this paper show, in deficient overall governance environments central banks tend to have independent investment committees because the independent committee can help safeguard decisions made by a weak and/or politically influenced board. However, the literature and data are scant on whether a central bank with supervisory functions impacts decision-making on reserve management and policy. This question may be another line of future inquiry and research.