A. Inflation, Growth, and Unemployment

Let us define “macroeconomic suffering” in terms of a simple index: (Inflation + Unemployment) – (Economic Growth), which is a variation of “Arthur Okun´s immiserising index” (Clavijo, 2000a). Table 1 shows that average annual inflation hovered around 22– 24 percent over the years 1975–97, while falling to 10.6 percent during 1998–2002. Note that the later figure is similar to the average of 12 percent registered over the “golden years” of the Colombian economy (1967–74). Unfortunately, this rapid disinflation was produced by, first, a weakening of the domestic aggregate demand since 1997 and, secondly, by a structural deterioration of the Colombian economy due to high indebtedness and the aggravation of the internal conflict.

Table 1.

Inflation, Unemployment, and Growth in Colombia

Source: Banco de la República.

Table 1.

Inflation, Unemployment, and Growth in Colombia

Periods	Inflation (CPI-Average) (1)	Unemployment (Main Cities) (2)	Growth (Real-GDP) (3)	Index of Macroeconomic Suffering (4)=(1)+(2)-(3)
1967–1974	12.1	9.9	6.2	15.8
1975–1981	24.7	9.5	4.5	29.6
1982–1989	22.6	11.7	3.5	30.9
1990–1997	24.0	10.1	4.0	30.0
1998–2002	10.6	18.1	0.5	28.3

Source: Banco de la República.

View Table

Table 1.

Inflation, Unemployment, and Growth in Colombia

Periods	Inflation (CPI-Average) (1)	Unemployment (Main Cities) (2)	Growth (Real-GDP) (3)	Index of Macroeconomic Suffering (4)=(1)+(2)-(3)
1967–1974	12.1	9.9	6.2	15.8
1975–1981	24.7	9.5	4.5	29.6
1982–1989	22.6	11.7	3.5	30.9
1990–1997	24.0	10.1	4.0	30.0
1998–2002	10.6	18.1	0.5	28.3

Source: Banco de la República.

This situation exacerbated unemployment, which averaged 18 percent in urban areas in recent years (and 15 percent nation-wide). Different studies show that the nonaccelerating inflation rate of unemployment (NAIRU) for Colombia has been around 10 percent. Hence, as aggregate demand grew weaker, the unemployment rate escalated and produced what has been termed an “opportunistic disinflation” (Clavijo, 2000a). However, there have been several episodes of financial and wage disindexation in Colombia, which permit us to be relatively optimistic about being able to maintain one-digit-inflation whenever aggregate demand recovers in the near future. This was also the case of United Kingdom, Israel, and Ireland, among others (Haldane, 1999).

The sum of inflation and unemployment (Okun´s immiserising index) averaged 34 percent during 1975–97. In spite of the rapid disinflation, this sum only declined to 29 percent in the period 1998–2002. When adding the effect of economic growth, which on average was nil in recent years, we find that our “suffering index” has been stranded at 28–31 percent over the last three decades (see Table 1). Note that this figure is twice the average observed over 1967–74, when the economy experienced export-led growth.

In short, the aggregate macroeconomic performance of inflation, unemployment, and growth has been poor over the years 1998–2002. In spite of a rapid disinflation process, high unemployment and low growth account for a relatively high index of macroeconomic suffering. However, this episode of “opportunistic dis-inflation” should be taken as a chance for reducing financial and wage indexation, so that a recovery of aggregate demand could take place in an environment of stable and low inflation.

B. The EFF Program with the IMF, 1999–2002

Table 2 reports the main macrovariables under surveillance (some of them constituted performance-criteria) within the IMF program and their results. It can readily be seen that the consolidated fiscal deficit reported slight overperformance during the years 1999–2000 (in the amounts of 0.2–0.5 percent of GDP). There was also significant overperformance in the current account of the balance of payments (0.7–1.9 percent of GDP), due to the decline in local aggregate demand and improved terms of trade (with an increase of 20 percent over 1999–2000). Real growth, however, under performed (by 0.7 percent in 1999 and 0.3 percent in 2000).

Table 2.

Targets and Results under the Extended Fund Facility Program with Colombia

Sources: IMF (2002a,b); and Banco de la República.

Table 2.

Targets and Results under the Extended Fund Facility Program with Colombia

	1999	2000	2001	2002
	(In percent of GDP)
Consolidated fiscal deficit (-)
Target	-6.0	-3.6	-2.9	-2.6
Result	-5.5	-3.4	-3.2	-4.0
Overperformance (+)	+0.5	+0.2	-0.3	-1.4
External deficit (-)
Target	-1.3	-2.4	-1.8	-3.2
Result	0.6	0.5	-1.9	-1.7
Overperformance (+)	+0.7	+1.9	-0.1	+1.5
	(Annual percentage change)
Real growth
Target	-3.5	3.0	3.8	3.0
Result	-4.2	2.7	1.6	1.7
Overperformance (+)	-0.7	-0.3	-2.2	-1.3
Inflation
Target	15.0	10.0	8.0	6.0
Result	9.2	8.8	7.7	7.0
Overperformance (+):	+5.8	+1.2	+0.3	-1.0

Sources: IMF (2002a,b); and Banco de la República.

View Table

Table 2.

Targets and Results under the Extended Fund Facility Program with Colombia

	1999	2000	2001	2002
	(In percent of GDP)
Consolidated fiscal deficit (-)
Target	-6.0	-3.6	-2.9	-2.6
Result	-5.5	-3.4	-3.2	-4.0
Overperformance (+)	+0.5	+0.2	-0.3	-1.4
External deficit (-)
Target	-1.3	-2.4	-1.8	-3.2
Result	0.6	0.5	-1.9	-1.7
Overperformance (+)	+0.7	+1.9	-0.1	+1.5
	(Annual percentage change)
Real growth
Target	-3.5	3.0	3.8	3.0
Result	-4.2	2.7	1.6	1.7
Overperformance (+)	-0.7	-0.3	-2.2	-1.3
Inflation
Target	15.0	10.0	8.0	6.0
Result	9.2	8.8	7.7	7.0
Overperformance (+):	+5.8	+1.2	+0.3	-1.0

Sources: IMF (2002a,b); and Banco de la República.

With respect to inflation, the 1999 target of 15 percent was over performed in as much as 5.8 percentage points and the 2000 target of 10 percent by 1.2 percentage points. Then, the disinflation was faster than expected by almost 7 percentage points in two years or 47 percent of the original 1999 target.

As mentioned, this overperformance of 1999–2000 is explained mainly by the weaknesses in local aggregate demand. In fact, note that inflation was brought down by 1.1 percent during 2001 (from 8.8 percent down to 7.7 percent), while growth descended two-full percentage points from the expected value (from 3.8 percent down to 1.8 percent). During 2002 the 6 percent target was missed by a 1 percentage point due mainly to weather factors that increased crop-prices temporarily. Nevertheless, nonfood inflation closed at 5.3 percent, below the target for headline inflation.

Since single-digit inflation targets imply higher outcome uncertainty in presence of supply shocks, there now exists an advantage in moving from point-targets to range-targets. In fact, it had been agreed with the IMF that deviations of +/- 2 percent from the inflation target would be allowed to accommodate such shocks. BR had also set up a range target of 5– 6 percent for 2003 and 3.5–5.5 percent for 2004, announced with the explicit purpose of anchoring inflation expectations. Adopting range targets will certainly have important policy implications, both at the level of the technical staff and at Board decisions.

Summarizing, the Board of Directors of the BR has moved from setting inflation point targets during the period 1991–2002 to setting range targets for the years 2003 and 2004. This is compatible with the framework of inflation targeting and the operational ranges that have allowed deviations of +/- 2 percent since 2001, within the IMF program. Furthermore, since uncertainty increases as inflation converges to the long-term value of 3 percent perannum adopted by the BR, excessive disinflation, as occurred during 1999–2000, should be avoided.

A. Elements of the New Monetary Policy

Colombia’s formal adoption of inflation targeting, since October 2000 (Banco de la República, 2000; Uribe, Gómez, and Vargas, 1999), has represented four main changes:

(a) Announcing multi-annual inflation targets, with the aim of leading expected inflation and expected nominal depreciation of the peso with respect to the dollar.

(b) Global assessment of the macroeconomic variables, with special attention being provided to the real sector and the solvency of the financial sector. In the former case, the credit transmission mechanism has been carefully assessed (through the repo rate) and in the latter case the liquidity access has been enhanced by linking the “lombard rates” with the “penalty rate” (in cases of requiring access to the discount window).

Put differently, new information besides monetary aggregates is being used extensively. In order to alter the monetary stance, the Board of the BR analyzes the output gap and assesses demand pressures. This information is used to forecast inflation 6, 12, and 18 months ahead. These results are further tested against possible trends of monetary aggregates, which are used as “lines of reference” for the monetary aggregates, not as intermediate targets. In my view, even these “lines of references” are of scant use for the BR watchers due to the continuous appearance of financial innovations, as occurred in the late 1980s (Clavijo, 1991). As a consequence, the exercise of forecasting monetary aggregates should be used only for internal purposes of the Central Bank and not as a tool for molding inflationary expectations.

(c) Signaling via interest rates. Since late 2000, the Board of the BR has also concentrated on transmitting the stance of the monetary policy through the central bank repo rate, which in turn guides the interbank rate.

The repo rate and the structure of lombard rates is now preferred as the instrument to communicate the desired stance of the monetary policy in Colombia. This procedure is very much in line with the current practice in the United States through the use of the Federal Funds Rate (FFR) and the discount window. In fact, starting in June 2000 the Fed ceased announcing any monetary aggregate forecast, due to the expiration of the Humphrey-Hawkins Act, dating from 1978 (Board of Governors (2000 p. 2) and Meyer (2001a p. 7–9)). Likewise, the Central Bank of Chile now concentrates on signaling the stance of the monetary policy through their repo rate (now in nominal terms), especially after adopting inflation targeting in 2000 and floating the peso against the dollar (García, et al, 2002).

It should be clarified that sometimes the stance of monetary policy could be altered without moving the repo rate, whenever the net asset position of the central bank is switched. For example, at the end of the year the BR usually is a net provider of resources, so the relevant rate for the market is the repo rate. However, in January the BR needs to contract the monetary base, so the relevant rate for the market turns out to be the reverse repo rate. On these occasions, it is crucial to transmit the proper message to the BR watchers.

The BR also uses a system of a ceiling lombard rate to establish the cost of accessing unlimited resources from the Central Bank, whenever the financial system experiences a fall of deposits or a floor lombard rate, to establish the highest remuneration for excess deposits left at the Central Bank. This is an alternative system to the discount window mechanism run by the Federal, which tends to diminish stigma problems (Hakkio and Sellon, 2000). In fact, the Fed announced in October 2002 the transformation of its discount window into a liquidity support mechanism by allowing solvent institutions to access resources at a modest penalty rate of about 100–150 basis points above the Federal Funds Rate (FFR), beginning in 2003.

This emphasis on transmitting the stance of the monetary policy via interest rates has permitted the establishment of semi-automatic mechanisms to compensate for changes in the composition of the monetary base. In early 2001, the BR created an Internal Operational Committee to deal with issues, very much in line with the daily practices of the New York Federal Reserve Bank, compensating, for instance, movements in Treasury accounts (Edwards, 1997 p. 863).

(d) Foreign Exchange Options: “put” to increase net international reserves (NIR) and “call” to decrease them. Since late 1999 the system of “put” options has permitted to increase NIR and to stabilize the permanent component of the monetary base. This mechanism has provided some discretion to monetary policy, making it more powerful in the short term. In 2001, the BR also announced the mechanism of “call” options in order to deal with eventual problems of excessive exchange rate depreciation that might spill over into inflation. Under such circumstances the Board would approve auctions in order to offer dollars to the financial system, with the aim of temporarily reducing the pace of nominal depreciation and maintaining the inflation target. The auction system adopted in Colombia comprises all public entities (including the Treasury). Finally, there exists a mechanism to control excessive volatility, set at +/- 4 percent of the 20-day moving average of the spot rate.

Table 3 illustrates the rules established under the option system. Under the “put” option modality, designed to increase NIR, the BR has offered to buy between US$30–200 million per month (about 1 to 5 percent of the market). These options are triggered when the spot exchange rate is more appreciated than the 20-day moving average of the spot rate. The cumulative amount exercised under these options was US$1,400 million over the period 1999–2002, representing 11 percent of the NIR at end-2002. Note that this value is lower than the 35 percent of the NIR accumulated in Mexico since 1995. Mexico suspended this mechanism back in August 2001, due to an apparent overaccumulation of NIR.

Table 3.

Foreign Exchange Options

(In millions of U.S. dollars; unless otherwise indicated)

Source: Author.

Table 3.

Foreign Exchange Options

(In millions of U.S. dollars; unless otherwise indicated)

I. “Put” Options to Buy Net International Reserves (NIR)
	Colombia (1999–2002)	Mexico (1995–2001)
Trigger rule	Spot < Spot(MA20Days)	Spot < Spot(MA20Days)
Amount offered per auction	30–200	250
Cumulative amount purchased	1,400	12,000
NIR	10,840	34,000
Amount purchased/NIR (percent)	11.3	35.0
NIR/amortization due (percent)	1.0	1.2
II. “Call” Options to Sell NIR
	Colombia (1999–2003)	Mexico (1995–2001)
Trigger rule	Spot > Spot(MA20Days)	--
Amount offered per auction	200 (or up to 1 billion)	--
III. Options to Control Volatility
	Colombia (1999–2002)	Mexico (1995–2001)
Trigger rule (percent)	Spot deviates by more than 4 percent from Spot(MA20Days)	Spot 2 percent more depreciated than Spot t-1
Amount offered per auction	180	200
Cumulative amount purchased	414	1,950
Exercised options/NIR (percent)	3.8	5.7

Source: Author.

View Table

Table 3.

Foreign Exchange Options

(In millions of U.S. dollars; unless otherwise indicated)

I. “Put” Options to Buy Net International Reserves (NIR)
	Colombia (1999–2002)	Mexico (1995–2001)
Trigger rule	Spot < Spot(MA20Days)	Spot < Spot(MA20Days)
Amount offered per auction	30–200	250
Cumulative amount purchased	1,400	12,000
NIR	10,840	34,000
Amount purchased/NIR (percent)	11.3	35.0
NIR/amortization due (percent)	1.0	1.2
II. “Call” Options to Sell NIR
	Colombia (1999–2003)	Mexico (1995–2001)
Trigger rule	Spot > Spot(MA20Days)	--
Amount offered per auction	200 (or up to 1 billion)	--
III. Options to Control Volatility
	Colombia (1999–2002)	Mexico (1995–2001)
Trigger rule (percent)	Spot deviates by more than 4 percent from Spot(MA20Days)	Spot 2 percent more depreciated than Spot t-1
Amount offered per auction	180	200
Cumulative amount purchased	414	1,950
Exercised options/NIR (percent)	3.8	5.7

Source: Author.

The “call” options system was first used in February of 2003, when the exchange rate was depreciating by nearly 30 percent annually, generating problems of pass-through that threatened the inflation target range of 5–6 percent for 2003. The simple announcement diminished the expected rate of nominal depreciation significantly. The amount offered for sale by the BR was US$200 million in the first month, out of a package of up to US$1 billion, consistent with the NIR targets established under the IMF-program. The options could be exercised when the spot exchange rate was more depreciated than the 20-day limit-moving around the spot rate.

The mechanism for controlling volatility was first used in 2001, but triggering conditions were not satisfied at that time. In 2002, the system was activated again, as volatility surpassed the 4 percent limit. On this occasion, the triggering conditions permitted the financial system to buy US$414 million from the BR (in several sessions) or about 3.8 percent of the NIR at end-2002.

In spite of the gains in the efficacy of monetary policy in Colombia, there exists a high degree of fiscal dependency. With structural fiscal deficits running at 3 percent of GDP, the pressure to issue local public debt is high, driving up real domestic interest rates. Furthermore, the financial sector portfolio has increased mainly through more investments in local treasury securities, which yielded 8 percent in real terms with no risk. Fortunately, during 2002 this condition began to change as financial credit grew in real terms for the first time in almost three years.

It is then crucial to break down this fiscal dependency to help credit flow to private investment. Otherwise, monetary policy will be facing serious impediments and growth will be hampered. The fiscal agenda has been well identified since the launch of the IMF program back in 1999, and luckily for Colombia great progress was made in late 2002 by approving the second generation of the pension reform, the labor reform, the state downsizing law, and a tax reform that imposed universal coverage for the VAT. The fiscal responsibility law was approved in 2003.

B. Central Bank’s Signaling: From “Quantities” (1991–94) to “Prices” (1999–2002)

The operational practices of the BR in the early 1990s followed the model of the Bundesbank: monetary aggregates were set as intermediate targets with the aim of reaching a desired combination of inflation and growth, in line with the so-called “monetary nominalism” (Clavijo, 2000a).

By contrast, the Fed began to switch, as early as October of 1982, towards signaling the stance of monetary policy through their interest rate instruments, namely, the federal funds rate (FFR) and the discount window (DW) (Hafer (2001)). Furthermore, and in line with this practice, the Fed model has been correctly characterized as one that has evolved towards a “disguised inflation targeting” by the mid-1990s, disregarding almost completely the targeting of monetary aggregates (Mankiw, 2001 p. 51).

During the period 1995–99, the BR altered significantly its operational procedures and moved towards interest rates as the main instrument of monetary policy. Simultaneously, the exchange rate system was allowed more flexibility, by turning the “crawling peg” system into a “crawling band” system, ending finally with the flotation of the peso in September 1999. Turbulence and contagion stemming from the Asian, Russian, and Brazilian crises ended almost three decades of fixing the exchange rate in Colombia, generating a debate about the benefits of having moved at that speed towards flotation (Hernández and Florez, 2000; Clavijo, 2000a; Villar, 2001; Fernández, 2001).

The Board of the BR felt that most of the conditions for adopting inflation targeting were in place in late 2000 and moved accordingly. In my view, there has been a breakthrough in the way monetary policy is being conducted after floating the peso, not simply an enhancement of procedures (Hernández and Tolosa, 2001 p. 27). To be sure, the exogeneity of the monetary base has permitted not only to increase the permanent component of the base but to provide independence for the instrument of the interest rate. Now the task is to consolidate this trinity framework in which monetary policy can work well in the long run, under a: (i) a flexible exchange rate system; (ii) an inflation target scheme; and (iii) a monetary policy rule based on signaling the stance through the repo interest rate of the BR.

C. The Transmission Mechanism of Monetary Policy

A. Objectives of Monetary Policy

In the case of Colombia, the 1991 Constitution defined the central bank as an autonomous institution in charge of fighting inflation, but acting in conjunction with the agenda of the Executive Branch. Indirectly, then, the BR has to take into account the government’s goals of promoting growth and generating employment. In this sense, the BR mandate is “hierarchical:” first, control inflation, but while taking into account the executive’s goals of higher growth and employment (Table 4). Put differently, Colombia’s central bank is not fully independent, but it is not subordinated to the government (Hernández, 1997 p. 87).

Table 4.

Banco de la República and Fed: A Brief Comparison

Source: Banco de la República.

¹FOMC denotes the Federal Open Market Committee.

Table 4.

Banco de la República and Fed: A Brief Comparison

			Banco de la República (Colombia) BR	Federal Reserve Bank (U.S.A.) Fed
Objectives			Hierarchical: Inflation Control and Coordination with Government Macro Policies	Dual: Inflation Control and Generation of Employment
Board members			Seven (Including the Minister of Finance)	Twelve for the FOMC 1/ Seven for the “Discount Window”
Strategy			“Inflation Targeting” (Explicit)	“Inflation Targeting” (Implicit)
Monetary instruments
	Central: Reference rates		Repo and Reverse Repo	Repo (Fed Funds Rate)
		Limiting rates	Lombard Rates-Discount Window	Discount Window
	Secondary: Aggregates		Monetary Corridors/ Reference Lines (announced)	Banking Reserves (un-announced)
	Support: Treasury		Semi-Automatic	Automatic
Exchange rates:
	Regime		Crawling Bands/Flotation	Flotation
	Instruments		Options: “puts” and “calls”	Intervention thru Treasury
Operational procedures
	Board Meetings:		Weekly (Various Subjects) Monthly (Inflation)	Every Month and a half (Inflation; Eight Times a year)
	Board announcements		Immediate	Immediate
	Board minutes:		By Petition (referring to Resolutions)	Two Months (Excl. Sensitive Topics)
	Transcriptions:		Confidential (for 3 years, if Resolutions)	Confidential (for 5 years)
	Reports:
		To public	Monthly	Every Month and a half (Eight per year)
		To Congress	Per Term	Per Term
	Role of staff		Active (recommends)	Passive (without recommendations)
	Econometric models		Simple model of transmission mechanism	Integrated National/International model

Source: Banco de la República.

¹FOMC denotes the Federal Open Market Committee.

View Table

Table 4.

Banco de la República and Fed: A Brief Comparison

			Banco de la República (Colombia) BR	Federal Reserve Bank (U.S.A.) Fed
Objectives			Hierarchical: Inflation Control and Coordination with Government Macro Policies	Dual: Inflation Control and Generation of Employment
Board members			Seven (Including the Minister of Finance)	Twelve for the FOMC 1/ Seven for the “Discount Window”
Strategy			“Inflation Targeting” (Explicit)	“Inflation Targeting” (Implicit)
Monetary instruments
	Central: Reference rates		Repo and Reverse Repo	Repo (Fed Funds Rate)
		Limiting rates	Lombard Rates-Discount Window	Discount Window
	Secondary: Aggregates		Monetary Corridors/ Reference Lines (announced)	Banking Reserves (un-announced)
	Support: Treasury		Semi-Automatic	Automatic
Exchange rates:
	Regime		Crawling Bands/Flotation	Flotation
	Instruments		Options: “puts” and “calls”	Intervention thru Treasury
Operational procedures
	Board Meetings:		Weekly (Various Subjects) Monthly (Inflation)	Every Month and a half (Inflation; Eight Times a year)
	Board announcements		Immediate	Immediate
	Board minutes:		By Petition (referring to Resolutions)	Two Months (Excl. Sensitive Topics)
	Transcriptions:		Confidential (for 3 years, if Resolutions)	Confidential (for 5 years)
	Reports:
		To public	Monthly	Every Month and a half (Eight per year)
		To Congress	Per Term	Per Term
	Role of staff		Active (recommends)	Passive (without recommendations)
	Econometric models		Simple model of transmission mechanism	Integrated National/International model

Source: Banco de la República.

¹FOMC denotes the Federal Open Market Committee.

In the case of the Fed, the central bank’s objectives are dual in terms of controlling inflation and promoting employment, especially after the 1946 and 1977 reforms (Meyer, 2001b). Furthermore, the Humphrey-Hawkins Act of 1978 included the promotion of economic growth as one of the objectives.

It is then clear that the BR and the Fed have only partial “political independence” (or so-called “instrument independence”) but in neither case has the objective of controlling inflation been set aside from the events of the real sector (Grilli, Masciandaro, and Tobellini, 1991; Walsh, 1993). To this respect, it is useful to recall the statement of the New York Fed Chairman: “Central banks neither can nor should be fully independent of government, since it is governments—and not central banks—that hold final responsibility for the economic and financial policy of the country. Nevertheless, some degree of central bank independence is critical” (McDonough, 1999, p. 5).

B. Board Members

The issue of “one main objective” (fighting inflation), but mediated by the principle of “coordination” between the Minister of Finance and the Central Bank, ends up being consistent with the mixed structure of the Board of Directors of the BR, which includes seven members:

• The Minister of Finance (the chairman),

• Five full-time Directors, chosen by the President, of which two may be replaced every four years, with a maximum tenure of 12 years in the post, and

• The General Manager, who is elected by the Directors for periods of four years, up to a maximum of 12 years in the post.

Each member has one vote, with no veto power by the Chairman or tie-breaking vote by the General Manager. The Head of the Office of Bank Supervision, a governmental body with some autonomy, has voice but no vote in the Board meetings.

It has been claimed that the Minister of Finance has special power in that he chairs the Board (Alesina, Carrasquilla, and Steiner, 2001). In practice, his unique edge stems from the fact that the Board cannot be convened without his participation (or that of his Deputy, when the Minister cannot attend). However, there are clear provisions to avoid Board meeting delays for more than two weeks and any member of the Board can call for extraordinary meetings at any moment to address special issues. Board members decide the agenda and the staffs of the Central Banks and the Ministry work jointly on the technical issues (Clavijo, 2000b).

By contrast, decision making at the Fed has two areas. The first one deals with the Federal Funds Rate (FFR) at the level of the Federal Open Market Committee (FOMC), where votes are cast by the seven independent Governors (including the Chairman and the Vice-Chairman) and the five (out of 12) Presidents of the Federal Reserve Banks (where New York has a permanent seat). The second area is related to the discount window, which is usually set slightly below the FFR, by the seven Governors of the Fed. This will change slightly beginning in 2003 as the discount window transforms into a more agile mechanism for supplying liquidity into the system, although at a penalty rate with respect to the FFR.

C. Anti-Inflationary Policy

The BR gradually moved towards inflation targeting over the years 1997–99 and formally announced it in October 2000. Interestingly, the Fed has had a peculiar system since the early 1990s in which explicit targets are not announced. The Fed operates with a “disguised inflation target” of around 3 percent per year, making careful analysis about trends of “core inflation” (Mankiw, 2001). The Fed has no explicit commitments to the Executive Branch and the Budget does not present a guide for setting inflation targets. In general, there has been good communication between the Fed, the government, and Congress, although tension has built up on certain occasions (see Woodward, 2000).

D. Monetary and Exchange Rate Instruments

As mentioned, the BR and the Fed have “instrumental independence.” The BR uses a system of a central repo rate to lead the interbank rate, where the lombard rates are used to control overflows of the financial system. The Fed uses the FFR and the discount window. In the case of BR there are semi-automatic mechanisms for compensating movements of the Treasury accounts, while in the Fed this occurs automatically on a daily basis.

Both central banks allow the flotation of their currencies, but in the case of Colombia the system is operated directly by the central bank’s staff (including the system of “options,” already explained), while in the case of the United States, the Treasury dictates the stance of the exchange rate policy, in conjunction with the Fed.

E. Operational Issues

One important feature of inflation targeting is the dissemination of broad and prompt information to the markets to operate in a transparent environment (Debelle, et al 1998). Board meetings at the BR to analyze compliance with price targets and to decide on the stance of monetary policy take place once a month, while at the Fed these occur every month and a half. In both cases actions are announced immediately to the media (see Table 3).

However, minutes of the board are known only about two months later in the case of the Fed, excluding sensitive information. In Colombia, such minutes are provided only under certain conditions. Furthermore, transcriptions are provided on the relevant areas after three years in Colombia, while in the United States they remain confidential for up to five years. Both the BR and the Fed provide regular information about the state of the economy and report formally to Congress twice a year. (See Urrutia (2001) for details on Colombia and Lindsey (2000) for details on the United States.)

The possible effect of changes in the interest rates are analyzed by using econometric models. In the case of the BR they are based on the so-called model of the inflation transmission mechanism (Gómez and Julio, 2001; Charry and Gómez, 2001). This system is improved constantly to better capture the dynamics of international markets and the pass through of the exchange rate to prices. In the case of the Fed, they include sophisticated macro-models of both the local and the international markets. One important difference is the role of the staff: it is very active in the case of BR, with explicit recommendations on policy actions to the Board, while in the case of the Fed it has a passive attitude, avoiding a preference for a particular stance of monetary policy.

A. Reading the Stance of Monetary Policy

The relationship between the nominal interest rate and the rate of growth of nominal GDP is sometimes used to decipher the stance of monetary policy. If the nominal interest rate exceeds the rate of growth of nominal GDP, one could infer that the position of the central bank is tight. In the case of Colombia, the years 1996 and again 1998–99 showed such a tight condition. However, during years 2001–2002 monetary policy was neutral, according to this measure.

Alternatively, one could judge the stance of monetary policy by directly computing the real cost of accessing resources from the Central Bank. Figure 1 illustrates the evolution of the repo rate of the BR and the Federal Funds Rate (FFR), after discounting the effect of inflation. During the years 1995–98 the repo rate was reduced from 20 percent to 8 percent, showing clearly a relaxation of monetary conditions in Colombia. However, the contagion from the Asian-Brazilian crises forced the BR to increase it to 12 percent in mid-1998. After managing the crisis, by recurring to the flotation of the peso, the BR continued to relax monetary policy in light of clear weaknesses of the local aggregate demand. The real repo-rate has been close to zero since late 2001. In the case of the United States, the nominal FFR fluctuated between 4–6 percent and finally declined to 1.25 percent by end-2002. In real terms, the FFR hovered around 2–4 percent during 1998–2000, rapidly falling to negative figures since late 2001, when the impact of recession was evident, aggravated by the terrorist attacks of September 11.

Real Interest Rates: Repo-Rate of the BR and Federal Funds Rate (FFR), 1995-2002

Citation: IMF Working Papers 2004, 166; 10.5089/9781451858136.001.A001

Source: Banco de la República.

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Real Interest Rates: Repo-Rate of the BR and Federal Funds Rate (FFR), 1995-2002

Citation: IMF Working Papers 2004, 166; 10.5089/9781451858136.001.A001

Source: Banco de la República.

• Download Figure
• Download figure as PowerPoint slide

Real Interest Rates: Repo-Rate of the BR and Federal Funds Rate (FFR), 1995-2002

Citation: IMF Working Papers 2004, 166; 10.5089/9781451858136.001.A001

Source: Banco de la República.

• Download Figure
• Download figure as PowerPoint slide

B. Taylor Rules

Economic agents follow the actions of the central bank and try to come out with a “reaction function” of the central bank to economic events. Central bank reaction functions that are set up (implicit or explicitly) in terms of policy interest rates are known as Taylor Rules (1993). In fact, in the case of the federal funds rate (FFR), it has been shown that such a rule follows the criteria of:

1. presenting a premium with respect to the long-term real interest rate;

2. increasing whenever the inflation surpasses the long-term target set up by the FOMC; and

3. increasing whenever real output surpasses potential output.

Table 5 shows a family of theoretical Taylor rules and their empirical counterparts for the FED. The basic Taylor rule postulated that the long-term real interest rate and long-term inflation rate converged to 2 percent, meaning (r* = π* = 2 percent). Then, short-term deviations of inflation with respect to such a value or deviations of output with respect to potential output, where y = Y – Y*, called for increases of the FFR of about 1.5 percent and 0.5 percent, respectively.

Table 5.

Central Bank Reaction Functions: Case of the Fed

Sources: Taylor (1993) and Rudebusch (2001).

Table 5.

Central Bank Reaction Functions: Case of the Fed

Interest Rate Rules		Theoretical Models	Estimations
1.	Basic Taylor rule	i = r* - 0.5π* + 1.5π + 0.5y
2.	Generalized Taylor rule	i = k + gππ + gyy	i = 0.63 + 1.7π + 0.8y + ε
3.	Optimal Taylor rule		i = 2.21 + 2.8π + 1.6y + ε
4.	Optimal dynamic	i = (1-ρ)(k + gππ + gyy) + ρi−1	i = 2.21 + 2.8π + 1.8y + ε
5.	Optimal lagged	i = k + gππ + gyy-1	i = 2.21 + 2.5π + 1.6y + ε

Sources: Taylor (1993) and Rudebusch (2001).

View Table

Table 5.

Central Bank Reaction Functions: Case of the Fed

Interest Rate Rules		Theoretical Models	Estimations
1.	Basic Taylor rule	i = r* - 0.5π* + 1.5π + 0.5y
2.	Generalized Taylor rule	i = k + gππ + gyy	i = 0.63 + 1.7π + 0.8y + ε
3.	Optimal Taylor rule		i = 2.21 + 2.8π + 1.6y + ε
4.	Optimal dynamic	i = (1-ρ)(k + gππ + gyy) + ρi−1	i = 2.21 + 2.8π + 1.8y + ε
5.	Optimal lagged	i = k + gππ + gyy-1	i = 2.21 + 2.5π + 1.6y + ε

Sources: Taylor (1993) and Rudebusch (2001).

The second line of Table 5 shows a generalized Taylor rule, where k ≡ r* - (g_π - 1) π*. Estimations for the 1987–96 period indicate that k = 0.63 and that inflation and product gaps closely follow the theoretical values mentioned above, since g _π = 1.78 and g _y = 0.82. However, when comparing these parameters with the implicit values derived from a Fed reaction function that contemplated dynamic effects (lines 3–5), it is clear that “observed” values represented “slow” actions, as if the Fed were too timid (Rudebusch, 2001). One possible explanation arises from concerns regarding excess volatility caused by Fed actions. This argument lends support for central bankers acting firmly but in a slow fashion, since changes of direction can harm economic stability.

C. Case of the Banco de la República

Table 6 illustrates variations of Taylor rules for the case of emerging markets. Equation 1 postulates that the central bank could also react to changes in the real exchange rate (q_t) and to its dynamics (q_t-1). As mentioned by Taylor (2001 p. 266), this does not make much sense in cases where alterations in q_t correspond to changes in productivity. Furthermore, temporary changes in the real exchange rate do not affect the long-term solution of such a rule, where f > 1 and g_y > 0, since these changes would show up in either inflation or product alterations. Put differently, the dynamics of a Taylor rule for an open economy are consistent with setting the values h _o = h ₁ = 0. However, in the case of the European Central Bank, the estimated net effect of a real depreciation of 10 percent of the Euro with respect to the dollar could require an increase of an additional 1 percent in the short-term interest rate of the European central bank.

Table 6.

Central Bank Reaction Functions: Case of Banco de la República

(Dependent Variable: Interbank Interest Rate)

Source: Taylor (2001) and our computations. Values in parenthesis show the significance level of the t-statistic and, for the t and t-2 values, correspond to the Chi-Square Statistic.

Table 6.

Central Bank Reaction Functions: Case of Banco de la República

(Dependent Variable: Interbank Interest Rate)

Interest Rate Rules		Theoretical Models	Estimations
1.	Taylor rule in an open economy: Targeting real exchange rate	i = fππ + gyy + h0qt + h1qt-1
			Period 1989–2002 (Quarterly):
2.	Open economy: Targeting monetary aggregates		A. Contemporaneous Values:
		i = r* + gm(M - M*) + gi(i* + e) + gyy	${i}=\underset{(99\%)}{23.9}-\underset{(40\%)}{0.04}({M}-{M}*)+\underset{(30\%)}{0.05}({i}*+{e})+\underset{(85\%)}{1.67}{y}$
			R2 = 0.72; Dw = 2.1; AR(1) = 0.84
			B. Lagged and Contemporaneous Values:
			$\begin{array}{cccc}{i}=\mathrm{16,6}-& {\displaystyle \sum _{{j}=0}^{2}0.03{({M}-{M}*)}_{{t}-{j}}}& +{\displaystyle \sum _{{j}=0}^{2}0.39{({i}*+{e})}_{{t}-{j}}}& +{\displaystyle \sum _{{j}=0}^{2}3.69{{y}}_{{t}-{j}}}\\ (99\%)& (1\%)& (99\%)& (99\%)\end{array}$
			R2 = 0.77; Dw = 1.95; AR(1) = 0.45
			Period 1998–2002 (Monthly):
3.	Open economy: Inflation targeting		C. Contemporaneous Values:
		i = r* + gπ(π - π*) + gi(i* + e) + gyy	${i}=\underset{(20\%)}{4.0}+\underset{(78\%)}{0.64}({\pi }-{\pi }*)+\underset{(66\%)}{0.13}({i}*+{e})-\underset{(74\%)}{0.11{y}}$
			R2 = 0.86; Dw = 2.29; AR(1) = 0.53; AR(2) = 0.41
			D. Lagged and Contemporaneous Values:
			$\begin{array}{cccc}{i}=5.3-& {\displaystyle \sum _{{j}=0}^{2}0.69{({\pi }-{\pi }*)}_{{t}-{j}}}& +{\displaystyle \sum _{{j}=0}^{2}0.18{({i}*+{e})}_{{t}-{j}}}& +{\displaystyle \sum _{{j}=0}^{2}0.09{{y}}_{{t}-{j}}}\\ (48\%)& (47\%)& (27\%)& (7\%)\end{array}$
			R2 = 0.84; Dw = 2.31; AR(1) = 0.46; AR(2) = 0.44

Source: Taylor (2001) and our computations. Values in parenthesis show the significance level of the t-statistic and, for the t and t-2 values, correspond to the Chi-Square Statistic.

View Table

Table 6.

Central Bank Reaction Functions: Case of Banco de la República

(Dependent Variable: Interbank Interest Rate)

Interest Rate Rules		Theoretical Models	Estimations
1.	Taylor rule in an open economy: Targeting real exchange rate	i = fππ + gyy + h0qt + h1qt-1
			Period 1989–2002 (Quarterly):
2.	Open economy: Targeting monetary aggregates		A. Contemporaneous Values:
		i = r* + gm(M - M*) + gi(i* + e) + gyy	${i}=\underset{(99\%)}{23.9}-\underset{(40\%)}{0.04}({M}-{M}*)+\underset{(30\%)}{0.05}({i}*+{e})+\underset{(85\%)}{1.67}{y}$
			R2 = 0.72; Dw = 2.1; AR(1) = 0.84
			B. Lagged and Contemporaneous Values:
			$\begin{array}{cccc}{i}=\mathrm{16,6}-& {\displaystyle \sum _{{j}=0}^{2}0.03{({M}-{M}*)}_{{t}-{j}}}& +{\displaystyle \sum _{{j}=0}^{2}0.39{({i}*+{e})}_{{t}-{j}}}& +{\displaystyle \sum _{{j}=0}^{2}3.69{{y}}_{{t}-{j}}}\\ (99\%)& (1\%)& (99\%)& (99\%)\end{array}$
			R2 = 0.77; Dw = 1.95; AR(1) = 0.45
			Period 1998–2002 (Monthly):
3.	Open economy: Inflation targeting		C. Contemporaneous Values:
		i = r* + gπ(π - π*) + gi(i* + e) + gyy	${i}=\underset{(20\%)}{4.0}+\underset{(78\%)}{0.64}({\pi }-{\pi }*)+\underset{(66\%)}{0.13}({i}*+{e})-\underset{(74\%)}{0.11{y}}$
			R2 = 0.86; Dw = 2.29; AR(1) = 0.53; AR(2) = 0.41
			D. Lagged and Contemporaneous Values:
			$\begin{array}{cccc}{i}=5.3-& {\displaystyle \sum _{{j}=0}^{2}0.69{({\pi }-{\pi }*)}_{{t}-{j}}}& +{\displaystyle \sum _{{j}=0}^{2}0.18{({i}*+{e})}_{{t}-{j}}}& +{\displaystyle \sum _{{j}=0}^{2}0.09{{y}}_{{t}-{j}}}\\ (48\%)& (47\%)& (27\%)& (7\%)\end{array}$
			R2 = 0.84; Dw = 2.31; AR(1) = 0.46; AR(2) = 0.44

Source: Taylor (2001) and our computations. Values in parenthesis show the significance level of the t-statistic and, for the t and t-2 values, correspond to the Chi-Square Statistic.

The problem of postulating a Taylor rule that includes the real exchange rate is that it turns out to be of scant practical use, due to well-known problems of uncertainty regarding the long-term purchasing power parity. In our opinion, it is more relevant to introduce instead the (uncovered) interest rate parity condition, since capital flows drastically affect net international reserves (NIR) and, consequently, the composition and value of monetary aggregates and its relations with local interest rates. In equation 2 we adopt this approach, where the condition i < (i* + e) induces capital outflows (here i* stands for the external interest rate and “e” represents the expected rate of depreciation of the local currency with respect to the foreign currency).

We distinguish two periods: 1989–2002 (quarterly), characterized by the targeting of monetary aggregates and a fixed exchange rate system (crawling peg and crawling bands); and, 1998–2002 (monthly), a period of flotation of the peso, inflation targeting, and signaling through the repo rate of the BR.

Equation (2) postulates that, during the 1989–2002 period, the market interest rates increased whenever the monetary expansion surpassed the quantitative target of the money base, there existed negative disparity with respect to external financial assets, and/or a positive product gap. For the period 1989–94, the postulated dependent variable was either the market interest rate (known as DTF) or the interbank rate and, for the period 1995–2002, the dependent variable was either the interbank rate or the repo rate, when it was instituted as one of the main instruments. Best results were achieved when using the interbank rate, which are reported in Table 6.

The econometric results are not satisfactory when using just contemporaneous variables, since we could only explain about 72 percent of the variation of the interbank rate and mainly by the effect of autoregressive movements, with great persistence of the inflation rate. For instance, in case A of equation (2) in Table 6, only the output gap seems to have played a marginal role, while monetary or interest rate gaps were not significant.

It should be said that part of the econometric problems arise from attempting to measure such gaps. Monetary targets where changed rather frequently, including new definitions of such aggregates (as M1 or M3), although the definition of the monetary base prevailed for the period as a whole. The output gap is rather sensitive to the definition of the method of decomposition (first we used a Hodrick-Prescott filter, but decided to move to a gap computed through a structural model). In fact, the order of integration of the variables represent a challenge in terms of long-term interpretations as the gaps are I(0), while the interest rates are I(1).

However, when considering contemporaneous and lagged values, econometric estimates improved significantly. Case B in Table 6 shows that the (uncovered) interest rate gap of the last three quarters exert a significant effect of about 39 basic points over the interbank rate. Additionally, the lagged product gap was also significant and showed a 3.7 value with respect to the log difference between the observed and the potential real GDP. The monetary aggregates gap remained insignificant.

We expected better econometric results for the period 1998–2002 (monthly), thanks to the adoption of inflation targeting and the consolidation of the repo rate as the main instrument of monetary policy. However, it seems that the span of the historical experiment is still too short. In fact, case C in Table 6 shows that the effect of the inflation gap is higher than that of the monetary gap (0.64 vs. 0.03), but significant only at the margin (78 percent). Something similar occurs with the (uncovered) interest rate gap and the product gap, where the later is related only to the industrial production. In order to overcome these problems, we introduced lagged values, but better results were not achieved (see case D in Table 6).

In short, we have seen that it is rather premature to postulate the existence of a clear reaction function of the BR regarding the repo rate with respect to inflation and output gaps for the period 1998–2002. There are theoretical and practical reasons to expect that in the near future econometric estimates should improve, especially after taking into consideration the effect of the (uncovered) interest parity condition, which significantly alters monetary policy in emerging markets.

Another strategy which is worth exploring is the use of forecasting values to compute the gaps to which the monetary authorities might be reacting. However, our experience tells us that such forecasting values have been rather inconsistent within a short period of time, so their direct use in the reaction function of the BR has not been a practical option up to now. This is one feature of the debate that intends to distinguish between inflation targeting seen as “an instrument rule” from the one that postulates inflation targeting as a “targeting rule” (Svensson, 2002).

Finally, we should also remark that the current strategy of inflation targeting in Colombia is threatened by fiscal dependency. High fiscal deficits, hovering around 3 percent of GDP, drive real interest rates up, and the long-term stability of the central bank reaction function is not clear under such dependency. Furthermore, this fiscal dependency goes beyond seigniorage whenever capital markets depend critically on the stability of public bond prices (Friedman, 1986 p. 17). As emphasized by Woodford (2001 p. 671), the central bank would feel the pressure of the fiscal deficit, not through monetary issue, but through pressure to sustain the value of public bonds, as in the “Accord of 1940” between the Fed and the U.S. Treasury.

In the case of Colombia, the high level of public debt in relation to GDP, now about 55 percent, makes it imperative to further coordinate monetary and fiscal policies (Banco de la República, 2002; Clavijo, 2002). The central bank should enhance the mechanisms that allow the consolidation of inflation targeting (such as maintaining the flotation of the peso), while efforts continue to lower public debt in relation to GDP.