7 Czech Republic
- International Monetary Fund
- Published Date:
- April 2005
Developing a Sound Governance and Institutional Framework
Reserve management objectives and coordination
462. The functions and objectives of holding reserves determine the reserve management strategy of the Czech National Bank (CNB). There are two main functions of reserves. The first is to serve as a source of liquid funds for meeting known and potential foreign liabilities and obligations of the central bank. In a wider sense, the obligations of the CNB are (i) to meet foreign exchange intervention requirements; (ii) to meet foreign exchange payments of CNB’s clients (of which the government is the most dominant one); and (iii) as a “backup” for meeting balance of payments crises. The second function is to serve as a form of national wealth, although it is only the central bank that has access to reserves. Foreign exchange reserves are a by-product of the monetary and exchange rate policy of the central bank. Due to the exchange rate regime and other developments since 1993, foreign exchange inflows are being absorbed into reserves. The foreign exchange inflows in the form of foreign direct investment (FDI) or speculative capital put pressure on the domestic currency, and by reverse interventions, the foreign exchange is accumulated in foreign exchange reserves of the CNB.
463. The general objective of reserve management is to ensure that reserves are positive, stable, and readily available for use. Positiveness and stability represent the risk limits, and availability for use ensures liquidity. This is a widely defined objective, and more detailed and concrete objectives are defined by the management of CNB depending on the conditions in the markets. However, even clear objectives are not easy to transform into measures of risk management such as investment horizon, duration, or the best currency composition.
464. From the inception of the Czech koruna in 1993 until 1997, it was pegged to a basket of two currencies, namely, the deutschemark and the U.S. dollar. It was easy to define the currency allocation in terms of the basket while at the same time avoid accounting losses. On a daily basis, the central bank bought (rarely sold) hard currencies. Each reserve currency was divided into two portfolios. One was the investment portfolio, which was fixed in size, and the other was the liquidity portfolio, which absorbed all inflows. The risks in each portfolio were managed separately with separate benchmarks primarily for duration and instruments. During the first three years of its existence, the central bank had accumulated substantial reserves with the level increasing by almost nine times. These inflows reduced the overall duration of the reserve to zero, thereby affecting returns when the yield curve was upward sloping. This situation arose because the focus was too much on the individual portfolios instead of on the entire reserves. There were no automatic procedures that would properly adjust the risk profile of reserves (the overall duration to development of balance of payment). This policy was changed in 1999 and today all the limits are set up first at the overall level, and the limits of individual portfolios are derived from the overall limits. Furthermore, the limits are expressed in relative terms (i.e., their absolute level is a function of the size of reserve) and the future development of reserve size is taken into consideration.
465. It is important to note that except for the central bank’s obligations and liabilities, the foreign reserves do not reflect the liabilities of any other entity. Even the central government’s position in foreign exchange is treated separately from the position of the central bank. If the government would like to either eliminate or hedge its foreign exchange position it must deal with the CNB. The central bank does not consider the government’s foreign exchange position while structuring its reserves. The reason is that it would be hard to differentiate between the position of the central government and positions of other institutions and companies financed from the central budget. For example in the Czech Republic, the government owned a company dealing with gas. The company was buying gas from abroad and distributing it here, meaning that this company’s assets were in Czech koruny while its liabilities were in U.S. dollars. This was exactly the opposite of the central bank balance sheet. If the dollar strengthened against the koruna, the central bank booked a profit and the company booked a loss. This loss had to be covered from the central budget. The profit of the central bank is by law transferred to the central budget. Both the central bank and the company have separate budgets and it could happen that both have the same position (if the company decides to hedge its foreign exchange position). The question is whether such a setup is enough to (i) have foreign reserves in the same structure as the company’s liabilities, and (ii) allow the company to avoid hedging its liabilities against foreign exchange risk just because the loss is compensated by the central bank’s profit. We believed that this is not a manageable situation.
466. The CNB, since its inception after the fall of the old central planned economy, has performed all the functions, including reserve management. In the Law of the CNB, international reserves management is defined as one of the main activities of the central bank. When carrying out this activity, the CNB acts independently of the government. The reserves are part of the balance sheet of the CNB. However, by the same law, all the property of the CNB, including the reserves, is the state property, which the CNB is mandated to manage. Therefore it is sometimes difficult to point out the legal owner of reserves. The CNB can freely deal with the reserves and the return on reserves is part of the CNB’s total performance. The bank’s net profit, after appropriation to certain funds, is transferred to the central government budget. The performance report that provides the details of the CNB’s profits and losses is submitted every year to the Lower House of the Czech parliament. The above description confirms the central bank’s independence from the government and parliament, but also points out that its operations are under certain supervision of the government and Parliament. This kind of supervision creates pressure on providing positive and stable return, indirectly influencing the risk profile of reserves.
467. The CNB Board, which is the highest decision-making level in the CNB, approves the basic strategy of reserve management and the permissible instruments. The strategy is defined by setting the currency allocation and maximum duration, and stipulating rules for credit and operational risk management and rules for portfolio management. The executive director for risk management sets the credit limits and the benchmarks (currency benchmark within 5 percent from the strategic currency allocation). The portfolio managers cannot deviate from the currency benchmark but can deviate from the duration benchmark up to ±20 percent only. The portfolio management rules permit deviation from the benchmark on credit quality, that is, investment in securities issued by issuers other than those included in the benchmark. Two separate departments—the Risk Management and Transactions Support Department and the Financial Markets Department—provide for execution and control of reserves management independently of each other.
Transparency and accountability
468. The CNB is not required under any regulation, including an internal one, to publish details on reserve management. However, it regularly publishes the level of reserves in the “Statistics” section of its website and, in addition, in response to published standards of the IMF, it decided to change the structure of the published information on reserve management in its Annual Report for 2001. The new structure provides more information on the results achieved in reserve management rather than on the market conditions as was the case until now. The chapter on reserve management now consists of absolute and relative performance and risk profile of reserves, including credit, interest, and currency risks. The CNB subscribes to the SDDS of the IMF and has been publishing its balance sheet since the last decade. Information on the basic structure of reserves and foreign liabilities of the CNB can therefore be accessed.
469. The foreign reserves management as a whole, including results, trading and settlement procedures, control and check systems, information systems, and distribution of responsibility are audited every year as part of the overall external audit of the CNB. The central bank does not carry out special audit of only the reserve management function. The auditor is selected by public tender, its term is five years, and both the CNB Board and the Minister of Finance must approve the selected auditor. The performance of the CNB is not audited by the highest auditing office but, as mentioned above, the performance is submitted to the Parliament. The Parliament can either agree or disagree with the report. If Parliament disagrees with the report, the central bank is requested to provide additional information and data.
470. There are two major information systems in use in the area of reserve management—one for the front office and the other for the back office. Both the systems have been developed internally. The decision to use internal capacity was made in 1996. Until that time, the portfolio managers, the risk managers, and the settlement staff used PC-based applications such as Excel or Lotus. It was a dilemma to choose between externally and internally built systems. The rationale for choosing the latter was that the CNB did not want to lose the option to incorporate in the future its procedures and methods for dealing, settlement, risk management, and accounting, which it would probably have to do if it chose the external system. Some financial limits also existed and influenced the final decision. Working with an external adviser, the basic content and structure of the system was designed in such a way that it is flexible to future internal changes in terms of organization structure, new products and instruments, new risk methods, changes of the accounting standards, and similar demands. Until now we have been successful in implementing functions like real-time distribution, deal entry, a database of historical prices, and a database of approved instruments. Within the next two years, we plan to implement the position keeping functionality and some of the risk measurement calculations, and connect the front office with the back office system. The advantages of an internal system are that it is flexible and friendly to changes, it is built under a controllable budget, and other aspects like backup or emergency installation are less difficult to do from the operational point of view. It is, however, unclear if these advantages are not offset by the fact that the system is still to be completed and the delay of four years limits the implementation of new risk measurement methods and introduction of new instruments. Because the project is in a very advanced stage, it was agreed that it should be finalized. But the planning and budgeting process and mainly the control of the time schedule and budget are based on the limited capacity of the CNB, which suffers a major handicap in IT projects compared to external and specialized firms.
471. In the field of the human resources, the position of the CNB does not differ a great deal from other central banks. On the position of portfolio manager, of which we have six positions, 24 men have changed during the last 11 years. A very similar situation exists in the area of risk management, IT support, and open market operation. The limited financial capacity of the CNB to pay competitive salaries has resulted in most of our staff being fresh graduates from universities. We would like to share an interesting and useful experiment that was organized by the CNB together with the World Bank in 1991. Through the public tender, 25 men were selected according to their knowledge of mathematics, statistics, and English. These individuals were offered work in some commercial banks and in the central bank, mostly in treasury areas. They had to undergo a three-year course on financial markets, instruments, and banking; several short seminars; and affiliations in commercial banks abroad. The course was financed by the central bank and the commercial banks. The participants of the course agreed to return the loan, in the event they left the course before completion or if they left the banks they had joined. Currently, many of the successful participants have reached the highest levels of management in the central bank, the commercial banks, and financial institutions in Czech Republic (e.g., vice governor of the CNB, executive director of risk management of the CNB, board member in Komercni banka and GE Capital Bank, etc.).
Establishing a Capacity to Assess and Manage Risk
472. The CNB started active reserve management in 1991. At that time bank deposits were the only instrument used for investment. Foreign exchange trading was used as an additional source of return. Risk management was “operated” by both the front office and back office. The back office provided the liquidity structure and the dealers managed liquidity. There also was a book of limits for deposits with approved banks but limits were “impressionistic” rather than based on any strict formula or rule. The interest rate risk, the foreign exchange risk, and the operational risk were not monitored. The reporting consisted of the gains and losses from foreign exchange trading and did not include mark-to-market valuation of term trades such as forwards. Six dealers and four back office operators performed the entire reserve management function. In 1992, the central bank bought its first security, the U.S. Treasury bill, and thereafter the risk management department was established. The main task of the department was liability management, credit risk management, and benchmarking. With the help of the World Bank and neighboring central banks (the Austrian central bank mainly) the new book of limits was created based on the financial strength of the banks. The formula calculating the limits was a function of the bank’s rating, capital, and assets. The approach is currently more sophisticated and the term of the investment is considered, as a distinction is made between the trading counterpart and the debtor.
Interest rate risk
473. The benchmarking process started in 1993 with simple benchmarks applied on each of the portfolios. They were combinations of general maturities such as one-week and three- month LIBOR indices for the liquidity portfolios, and two-, three-, and five-year benchmark government notes and bonds for the investment portfolios. The overall duration, calculated ex post, was around 1.5 years in the very beginning, though the rationale for choosing this duration is difficult to find out. This approach was changed in 1999 when the new benchmark representing the entire reserve and two sub-benchmarks for each reserve currency were set up. The benchmark is a composite of a general maturities index for the money market part and the external Merrill Lynch government index for the medium- and longer-term part of the yield curve. The most important fact is that the overall duration is derived from the objectives and targets of reserve management. Duration is established based on the requirement that the portfolio should not record a loss in any three-month period. For setting the target duration, historical time series of yields on the relevant financial markets are used. In addition to historical data, the methodology for setting duration takes into account the current situation in the financial markets. If short-term interest rates are higher, the higher is the interest rate risk that can be accepted. By historical statistical data, it is proved that by using this approach the benchmark performs better by 5–10 basis points. This type of reset of the benchmark is done quarterly and therefore has marginal impact on transaction costs. It is our experience that neither the internal nor the external (public) benchmark would help to decide what duration and currency composition is ideal.
474. The currency risk was until May 1997 relatively easy to manage. The koruna was fixed to a basket of U.S. dollars and deutschemark, and the currency allocation of reserves was identical to the basket. The U.S. dollar and deutschemark offered the two most liquid markets and absence of yield diversification was compensated by the fact that there was no risk of accounting losses when the reserves were valued in domestic currency. In May 1997, the CNB decided to float the koruna, and the basket as an anchor for reserve currency allocation ceased to exist. The deutschemark was declared as the intervention currency and thus emerged large foreign exchange risk because reserves were in both U.S. dollars and deutschemark whereas the koruna was floating. Taking into consideration, the currency of intervention, benefits of diversification, and the need to operate in liquid and efficient markets, it was decided to hold reserves only in U.S. dollars, deutschemark, and yen and the currency allocation decision was based on the structure of the balance of payments, that is, 65 percent in deutschemark, 30 percent in U.S. dollars and 5 percent in Japanese yen. After three years, due to problems with rating of Japan and its banking system, it was decided to hold reserves in U.S. dollars and euros. The allocation of reserves into these currencies takes into account various factors, an important one being investment diversification. The aim is to attain the most stable income possible given the exchange rate between the reserve currencies. When setting the ratio between the two currencies in the international reserves, the CNB analyzes the historical time series of the yields on U.S. and European markets and the EUR/US$ exchange rate. Other factors taken into consideration include the nature of the domestic foreign exchange market, where EUR/CZK is the most important and most traded currency pair. Based on these considerations, the currency composition was set at 73.4 percent for euros and 26.6 percent for U.S. dollars.
475. Credit risk issues can be divided into two groups: issues relating to the selection of the issuers of the financial instruments used for reserves management, and issues relating to the selection of business partners for the execution of reserves management transactions. The sole acceptable issuers are the governments and central banks of OECD countries as well as certain governments and international organizations (e.g., the World Bank) and selected banks from those countries. Moreover, maximum maturity is limited to 10 years for government bonds and 3–6 months for claims on banks (depending on the bank’s rating). The credit risk management is based on two sets of limits—one for banks and financial institutions as the CNB’s counterparties, and the other for banks and financial institutions as debtors of the CNB. In both the cases, the limits are calculated by using a formula that is a function of rating, capital, and size of assets. The minimum rating of banks as debtors is individual C, long-term A-, and minimum assets of US$10 billion. Any exception can be made only by the Board of the central bank in writing with a fixed termination.
476. Operational risk control is based on the split between responsibilities for execution and for the benchmarking and settlement at the highest managerial level. There are two departments covering the reserve management, and their executive directors report directly to the board. One is responsible for trading and execution, the second is responsible for risk management, settlement, nostro accounts management, and for IT support. The data regarding the execution are passed on from the front office to both risk management and the back office. The back office also provides data about the settlements to the risk management. Within the risk management department, all the data are compared and the final position of the bank is created every day. The back office monitors the credit limits before the trades are settled, while the risk management monitors limits at the end of the day. The director of risk management has the right to stop the settlement if unauthorized trades are submitted to the back office. Since the executive directors in charge of execution and control are different, it is difficult to manipulate the results.
Experience with external managers
477. The CNB has had quite a long experience with external managers. The experience has been expensive and rather negative. During the last six years there were six mandates. The first two were signed in 1996. The goals at that time were to test if the internal benchmarks were realistic and to benefit from transfer of know-how. However, the external managers’ performance was worse than that of the internal managers and the activity of external managers’ was almost five times lower than that of the internal managers. The mandates were, therefore, terminated in 1999. In 2000, another three mandates were signed. This time the target was to test new instruments and ways of foreign exchange risk management. One mandate was supposed to test Pfandbriefs with a maturity of up to five years; the second one U.S. agencies’ global notes; and the third one’s role was to replicate our benchmark’s basket of currencies of 35 percent in U.S. dollars and 65 percent euros and to invest on a hedged basis in 12 other major government markets. All managers ranked among the biggest firms and, except for one (FFTW), all managers were chosen by tender. The same reporting requirements were applied and the amounts under management were around US$100 million. Even here the experience was negative since all managers had sizable problems with reporting (in one case two CNB experts had to visit the manager and help with the reporting system) and their activity was, in general, much lower than the activity of the internal managers and therefore know-how transfer was very limited. Finally, the fees paid to the managers were higher than the relative performance. The benefits of having external managers were that it helped build contact with the asset management industry, provided the results of the tests, and confirmed the right approach to reporting. Despite the negative experience, we are preparing another set of mandates that would test for U.S. corporate bond portfolio management, mortgage-backed securities (MBS), and foreign exchange trading. These are the areas of limited capacity of the CNB.
478. The external managers’ portfolios are subject to the same performance measurement rules and are used to verify and assess certain procedures that could potentially also be used for internal reserves management. These externally managed portfolios constitute approximately 2.3 percent of the reserves.
Operations in efficient markets
479. The CNB currently invests its reserves in only euro- and U.S. dollar-denominated deposits and government bonds. Almost by definition these markets are considered as the most liquid and efficient. That is why we do not test and analyze the depth of these markets.
480. It is important to note that one of the weak points of reserve management is that the Czech law is different from the international one, mainly the U.S. and the U.K. law, according to which most of the agreements are governed. The difference means that there are not enough lawyers familiar with the international laws and there have been very few events where these two different types of laws have been tested in the court. In the event of some legal dispute, the uncertainty is high.
Dilemmas in Reserve Management
481. Finally, the following is a list of dilemmas that must be solved generally and that the CNB is currently working on. The first dilemma is currency allocation versus the fact that the reserve’s biggest foreign exchange risk comes from the definition of reserves (i.e., the reserves are in currency other than the domestic and accounting currency). The foreign currency risk in a floating exchange rate regime (meaning an unhedgeable position) is so large and hopefully the horizon so long that it is not clear if it makes sense to deal with currency allocation of reserves unless one takes the market view on the reserve currencies’ exchange rates against each other.
482. The second dilemma is how to treat the ratings. Once the ratings express the probability of companies to default and once the central bank bases the credit limits on rating, the question is should the central bank differentiate between ratings of the banks and corporates. Why is it that no central bank has a problem with depositing funds in an AA bank, while only a few central banks would purchase AAA corporate debt?
483. The other dilemma is the link between the benchmarks and their traditional functions. The benchmark usually serves as the set of risk constraints expressed in assets that can be bought in the market (portfolio of real bonds, index, etc.). Managing the portfolio against the benchmark can mean the effort to beat the benchmark or the effort to beat the market within the limits. Beating the benchmark means that even a wrong investment decision may be considered in a relative sense as a value addition, even if the “benchmark’s” investment performs worse. Beating the market means that the portfolio manager is able to exploit the market opportunity and by definition would perform better than the benchmark. Our external managers tried predicting the yields within certain horizons; they went short or long and took maximum duration positions, while our internal managers focused primarily on replicating the benchmark. The comparison of these two approaches applied on our benchmarks during the last six years is by far in favor of internal managers. So the question is at what level the asset managers should take the market view—when setting up the benchmark or when managing the portfolio?
484. There is also another dilemma linked to the benchmark. Today it becomes standard to measure the risks by statistical methods. We can either compare the duration of two “portfolios”—where one is the real portfolio and the second is the benchmark—or we can compare two VaRs—where one is the value at risk of the real portfolio and the second is just the limit number meaning how much we are prepared to lose. However, apart from the risk, benchmarks are used for performance measurement despite the fact that the benchmark may not provide the best investment alternative. Therefore one can eliminate the benchmark portfolio for risk measurement purposes and measure the performance against an “ex post” calculated number. The proposal is that every day we can for certain markets and certain instruments (like the U.S. Treasuries market) calculate the best investment opportunity. And this best investment opportunity would be declared as the benchmarked performance for the previous day. To avoid the situation where the benchmark consists of one bond, some minimum holdings would have to be applied. Also, calculating the best investment out of several hundreds of bonds might be technically difficult. Therefore the best investment calculation would be done on a narrow set of bonds (instruments) that would best represent the market moves.
485. If the above methods are put together, the new benchmark would consist of VaR limit and return of the best overnight (alternatively over a certain period) investment alternative and no “benchmark as portfolio” would be needed.