## _wp0615

## Source details

**Canonical URL:** [_wp0615](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2006/_wp0615.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2006/_wp0615.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2006/_wp0615.pdf.json)

---

### I. Introduction
- Financially repressed regimes characterized by suppressed prices lower the rate of real economic growth and retard development.
- Suppressed inflation: at existing wages and prices, aggregate demand for current output and labor services exceed aggregate supply; results from inability of wages and prices to adjust instantaneously (Barro and Grossman, 1974).
- Two causes of price/wage nonadjustment:
  - Effective legal constraints (price and wage restrictions).
  - Natural frictions in market mechanisms.
- Nominal income increases can act as a money illusion, increasing the inflationary overhang and spurring maladjustments that reduce supply.
- Theoretical foundation: disequilibrium macroeconomics (Muellbauer and Portes, 1978; Barro and Grossman, 1974) — too low price and nominal wage levels produce excess demand for consumables and labor; households reduce effective labor supply in response to increased real money balances and consumption constraints.
- Paper aim:
  - Explain the substantial decrease in velocity and increasing real money balances that led to a divergence between inflation and monetary expansion since late 2003 in Zimbabwe.
  - Contextual facts:
    - Real GDP declined by almost 30 percent from 1997 to 2003.
    - Inflation rose from about 20 percent in December 1997 to a peak of 623 percent in January 2004.
    - Agricultural production collapsed following violent implementation of fast-track land reform.
    - Official exchange rate held highly overvalued; declining exports and foreign financing led to sharp import restrictions and external arrears; investment fell sharply; shortages pervasive.
    - Real GDP fell by about 4 percent in 2004, while year-on-year inflation decelerated from 623 percent in January 2004 to around 130 percent at end-2004, despite money growth remaining at very high levels.
- Paper structure:
  - Section II: background on evolution of inflation and money aggregates in Zimbabwe.
  - Section III: analysis of demand for money since the late 1990s.
  - Section IV: discussion of other factors leading to divergence of inflation and money growth in the short run.
  - Section V: conclusion.

### II. Stylized facts (inflation, money growth, velocity, and interest rates)
- Inflation dynamics (12-month changes):
  - About 20 percent in December 1997.
  - Peak of 623 percent in January 2004.
  - Decelerated sharply from March 2004 to around 130 percent at end-2004.
- Broad money (M3) growth dynamics (12-month changes):
  - Over 400 percent at end-2003.
  - Some 130 percent by end-December 2004.
  - Broad money growth started decelerating only in July 2004.
- Policy interest rate:
  - Raised sharply in Q1 2004, reaching a peak at 5,242 percent in annualized basis in March 2004, then subsequently lowered as inflation declined.
  - Real interest rates remained at very high levels throughout 2004.
  - Reserve Bank of Zimbabwe (RBZ) continued a subsidized credit facility providing low-interest loans to selected borrowers despite high real rates.
- Velocity:
  - Velocity rose significantly in 2003 with sharp acceleration in inflation, then declined quickly in 2004 — the year inflation started to fall.
  - Velocity estimates derived from Fisher equation with π equal to year-on-year CPI change rebased to GDP deflator in 2000, M as broad money, V as velocity of broad money, and Y as real GDP.
  - Stabilization defined as the first month when 12-month inflation falls below 40 percent.
- Real money balances:
  - Real money balances started increasing in 2004 while inflation was still very high — at odds with conventional U-shaped pattern over stabilizations (initial decline then increase during remonetization).
- Exchange rate and real balances:
  - Despite depreciation of the parallel market exchange rate in 2004, demand for real balances increased.

### III. Exchange rate regime change and parallel market dynamics (2003)
- A managed foreign exchange tender system was introduced and there was a clamp down on the parallel market in 2003.
- The tender rate depreciated only moderately for the rest of 2003 despite a growing gap between demand and supply in the tender.
- The parallel market resurfaced with a rapidly depreciating exchange rate.
- Real money balances continued to increase despite continuous depreciation of the parallel exchange rate.

### IV. Model specification and data for money demand estimation
- Framework: Friedman’s model of demand for money with long-run log-linear specification (Equation (1) rewritten as Equation (2)).
- Data (monthly, seasonally unadjusted; sample period 1998:1-2004:12):
  - Dependent variable: Broad money, M3 (m) — real broad money.
  - Domestic consumer price index (p).
  - Real GDP (y) — monthly GDP data generated by cubic spline interpolation.
  - Three-month time deposit interest rate (R).
  - Parallel exchange rate (e) — source: World Currency Yearbook and Techfin.
- Transformations and interpretation:
  - Lowercase variables are logs; ∆ denotes first difference of natural logarithm.
  - R, ∆p, and ∆e measure returns on M3, goods, and U.S. dollars, respectively.
- Estimation method: cointegration system estimated using the Johansen procedure.
- Assumptions/caveats: households assumed in equilibrium following Portes and Winter (1978); unit-root test results considered tentative given short sample.

### V. Empirical findings: unit-root, cointegration, and cointegrating vector
- Unit-root (ADF) test statistics (levels; ADF test with constant; 5% Critical Value: -2.90; Sample period: 1998:1-2004:12; Lag orders selected by AIC):
  - m-p: -1.9141
  - y: -0.1912
  - R: -2.1792
  - ∆p: -2.6661
  - ∆e: -2.6263
- Cointegration tests:
  - Maximal eigenvalue and trace eigenvalue statistics indicate rejection of no cointegration in favor of one cointegrating relationship (reported table entries: 0109.130.0056.860.00; 142.270.0530.520.02; 221.750.1317.150.17; 314.610.1112.340.10; 42.260.132.260.13).
- Estimated cointegrating vector (Dependent Variable: Real Broad Money; Sample period: 1998:1-2004:12; Monthly data; VAR lags: 6; R2: 0.82):
  - Real GDP coefficient: 0.4127*** (Std. Err. 0.106)
  - Time-deposit interest rate coefficient: -0.0007*** (Std. Err. 0.000)
  - Inflation coefficient: -1.2817*** (Std. Err. 0.326)
  - Depreciation coefficient: -0.1214* (Std. Err. 0.093)
  - Seasonal dummies: yes
  - Constant: yes
  - Significance: *, **, and *** correspond to 10, 5 and 1 percent levels, respectively.
- Interpretation:
  - Negative and significant coefficients on inflation and depreciation imply goods and U.S. dollars are substitutes for money.
  - Income elasticity ~ 0.4127, statistically not different from 0.5 and consistent with Baumol-Tobin transactions demand.
  - Time-deposit interest rate coefficient very small, negative, and significant at 1 percent; possible explanations include composition of broad money (primarily interest-bearing deposits) and limited variation in interest rates.

### VI. Diagnostics, forecasts, and structural break evidence
- Diagnostic tests (AR 1-5 portmanteau, Normality skewness and kurtosis, ARCH 1-5, Hetero test):
  - Estimated residuals free of autocorrelation and heteroscedasticity.
  - Normality not rejected except for weak non-normality for real GDP and depreciation.
  - Recursive estimation shows coefficients are constant over most of the sample.
- Forecasts and structural breaks:
  - Within-sample forecasts indicate real money balances would not have been expected to increase by end-2004 as actually occurred.
  - Residuals from fitted money demand function reveal 2004 as an outlying year.
  - Chow tests find a single spike indicating a structural break in 2004 — the year when velocity deviated from expected path.
  - Residuals suggest most observed points lie near a true money demand function, with 2004 departing significantly.

### VII. Short-run influences on velocity and measurement issues
- Short-run influences cited:
  - Inertial component of inflation (lagged price response).
  - Changes in demand for money (inflationary expectations) affecting velocity.
  - Wage policies, exogenous import price increases, and changes in administered prices affecting domestic relative prices.
  - Possible improper measurement of variables (CPI, money).
- Potential understatement of official inflation:
  - CPI may understate actual inflation because:
    - goods traded in the parallel market are not included in the CPI index;
    - CPI includes controlled and monitored prices;
    - CPI uses outdated weights from 1997.
  - Kovanen (2003) estimate: if full price liberalization had taken place in April 2003 with administered prices adjusted to market levels, overall CPI would have adjusted upward by about 47 percent.
- Additional facts:
  - Domestic energy prices were allowed to remain low despite increases in prices of energy imports.
  - Parallel market premium widened from 13 percent in January 2004 to about 53 percent in December 2004.
  - Pass-through to actual market prices may have been rapid anecdotally, but measured CPI may not have fully captured these increases.
- Net effect: understatement of inflation during recorded disinflation in 2004 could help explain the dramatic fall in velocity.

### VIII. Conclusions and policy implications
- Main empirical conclusion:
  - Except for 2004, a stable demand for money as a function of parallel market exchange rate, inflation, and real output can be found for Zimbabwe.
  - A structural break occurred in 2004: despite depreciation of the parallel exchange rate in 2004, real money balances increased.
- On explaining 2004 behavior:
  - A rapid fall in expected inflation could explain the 2004 increase in real balances, but this is atypical given inflation inertia.
  - Repressed inflation and mismeasurement of inflation are identified as the most likely possibilities.
  - Authors find no evidence of sustained disequilibrium behavior that definitively confirms repressed inflation; final judgment requires a disequilibrium analysis.
- Policy guidance:
  - Coordinate monetary policy with management of the current price system.
  - In economies with price controls and external-sector distortions, changes in money supply not accompanied by changes in domestic prices and exchange rate may produce repressed market demands and balance of payments disequilibrium.
  - Concern over liquidity overhang built up in 2004 with M3 exceeding its reference.
  - Quoted policy lesson: "price controls do not get at the fundamental source of inflation—excessive expansion of the money supply—and eventually make the problem worse." Controls often result in shortages and bottled-up inflationary pressures that can explode.

*Source: _wp0615 - IMF Working Paper content (Zimbabwean authorities and IMF staff estimates; sample 1998:1-2004:12).*

### References..............................................................................................................

### _wp0615 - References..............................................................................................................

### I. INTRODUCTION
- Financially repressed regimes are characterized by suppressed prices, which lower the rate of real economic growth and retard the development process.
- Suppressed inflation: at existing wages and prices, aggregate demand for current output and labor services exceed aggregate supply; results from inability of wages and prices to adjust instantaneously (Barro and Grossman, 1974).
- Two causes of price/wage nonadjustment:
  - Effective legal constraints (price and wage restrictions).
  - Natural frictions in market mechanisms.
- In such regimes nominal income increases occur as a money illusion to minimize social discontent and provide work incentives, increasing the inflationary overhang (accumulated and unusable purchasing power) while spurring maladjustments and inequities that further reduce supply.
- Theoretical foundations draw on disequilibrium macroeconomics (Muellbauer and Portes, 1978; Barro and Grossman, 1974):
  - Too low price and nominal wage levels produce excess demand for consumables and labor, causing employment and output below general-market-clearing levels.
  - Households respond to increased real money balances and consumption constraints by reducing effective labor supply; output constrained by employment and the production function.
- Prior literature and measures:
  - Portes and Winter (1978): market approach supposing no excess demand in official economy in Poland (1953–1973).
  - Charemza and Ghatak (1990): shortages in official sector matter for money demand; under shortage households hold additional money balances to make purchases at higher black-market prices.
  - Feltenstein and Ha (1991): derived variable measure of repressed inflation in China based on monetary overhang (excess of consumer money holdings over nominal value of retail goods).
- Aim of paper:
  - Explore explanations for substantial decrease in velocity and increasing real money balances that led to a divergence between inflation and monetary expansion since late 2003 in Zimbabwe.
  - Context: real GDP declined by almost 30 percent from 1997 to 2003; inflation rose from about 20 percent in December 1997 to a peak of 623 percent in January 2004; agricultural production collapsed due to violent implementation of fast-track land reform.
  - Official exchange rate held at a highly overvalued level; declining exports and foreign financing led to sharp restrictions on imports and accumulation of external arrears; investment fell sharply; shortages pervasive.
  - Real GDP fell by about 4 percent in 2004, while year-on-year inflation decelerated from 623 percent in January 2004 to around 130 percent at end-2004, despite money growth remaining at very high levels.
- Paper structure:
  - Section II: background on evolution of inflation and money aggregates in Zimbabwe.
  - Section III: analysis of demand for money since the late 1990s.
  - Section IV: discussion of other factors leading to divergence of inflation and money growth in the short run.
  - Section V: conclusion.

### II. STYLIZED FACTS
- General observation:
  - Money growth and inflation paths started to diverge from late 2003.
- Key numeric facts and dynamics:
  - Inflation path:
    - About 20 percent in December 1997.
    - Peak of 623 percent in January 2004.
    - Decelerated sharply from March 2004 to around 130 percent at end-2004.
  - Broad money growth:
    - Started decelerating only in July 2004.
    - Over 400 percent at end-2003.
    - Some 130 percent by end-December 2004.
  - Note: all growth rates in the paper are 12-month changes.
- Contrast with typical stabilization episodes:
  - Typical experience: inflation inertia evident; inflation lags the decline in money growth because price/wage expectations are partly based on past inflation and react slowly to changes in money growth; credibility issues can delay inflation decline.
  - Zimbabwe: prices appear to have responded to factors other than just changes in monetary policy.
- Policy interest rate dynamics:
  - Policy interest rates raised sharply in Q1 2004, reaching a peak at 5,242 percent in annualized basis in March 2004, and subsequently lowered as inflation declined.
  - Real interest rates remained at very high levels throughout 2004.
  - Reserve Bank of Zimbabwe (RBZ) continued to operate a subsidized credit facility providing low-interest loans to selected borrowers, despite high real rates and overvalued official exchange rate pressure.
- Velocity:
  - Velocity declined in the same year inflation started to fall.
  - 2003: significant rise in velocity associated with sharp acceleration in inflation, but velocity declined quite quickly in 2004.
  - Typical pattern during stabilization: sharp increase in velocity in initial stages, followed by decline when stabilization achieved.
  - Velocity estimates: derived as an implied index from the path of money, prices, and real output following the standard Fisher equation with π equal to the year-on-year change in the consumer price index rebased to the GDP deflator in 2000, M to broad money, V to velocity of broad money and Y to real GDP.
  - Stabilization defined as the first month when 12-month inflation falls below 40 percent.
- Real money balances:
  - Zimbabwe’s real money balances started increasing in 2004 while inflation was still very high — at odds with conventional wisdom of a U-shaped pattern of real balances over a stabilization program.
  - Conventional U-shaped pattern: decline in real balances during initial high-but-declining inflation; increase in real balances during remonetization with relatively stable/moderate inflation; the increase phase typically smaller than the decline phase (not symmetric).
- Exchange rate and real balances:
  - Despite a depreciation of the parallel market exchange rate in 2004, demand for real balances increased.
- Figures and tables referenced (data sources: Zimbabwean authorities and IMF staff estimates):
  - Figures include:
    - Figure 1. Inflation and Money Growth since 1999
    - Figure 2. Velocity and Inflation, 1980-2004
    - Figure 3. Growth of Real Money Balances and Inflation, 1980-2004
    - Figure 4. Growth of Real Money Balances and Parallel Exchange Rate, 1999-2004
    - Figure 5. Recursive Estimation
    - Figure 6. Within-Sample Forecast
    - Figure 7. Chow Tests
  - Tables include:
    - Table 1. Unit-Root Tests
    - Table 2. Cointegrating Tests
    - Table 3. Cointegrating Vector
    - Table 4. Diagnostic Tests

*Source: _wp0615 - References..............................................................................................................*

### introduction of a managed foreign exchange tender system and a clamp down on the parallel

### introduction of a managed foreign exchange tender system and a clamp down on the parallel market, where the bulk of foreign exchange transactions was taking place in 2003

### Background: exchange rate regime change and parallel market dynamics
- A managed foreign exchange tender system was introduced and there was a clamp down on the parallel market in 2003.
- The tender rate depreciated only moderately for the rest of 2003 despite a growing gap between demand and supply in the tender.
- The parallel market resurfaced with a rapidly depreciating exchange rate.
- Real money balances continued to increase despite the continuous depreciation of the parallel exchange rate.

### Model specification and data used to estimate money demand
- Framework: Friedman’s model of demand for money with the long-run log-linear specification (Equation (1) rewritten as Equation (2)).
- Variables (monthly, seasonally unadjusted, sample period 1998:1-2004:12):
  - Broad money, M3 (m) — dependent variable (real broad money).
  - Domestic consumer price index (p).
  - Real GDP (y) — monthly GDP data generated by cubic spline interpolation.
  - Three-month time deposit interest rate (R).
  - Parallel exchange rate (e) — source: World Currency Yearbook and Techfin.
- Transformations:
  - Variables in lowercase are in logarithms.
  - ∆ denotes first difference of natural logarithm.
  - R, ∆p, and ∆e measure returns on M3, goods, and U.S. dollars, respectively.
- Estimation method: cointegration system estimated using the Johansen procedure.
- Assumptions and caveats:
  - Households assumed in equilibrium following Portes and Winter (1978).
  - Given short time period, unit-root test results should be considered highly tentative.

### Empirical findings: unit roots, cointegration, and estimated cointegrating vector
- Unit-root (ADF) tests (levels; ADF test with constant; 5% Critical Value: -2.90; Sample period: 1998:1-2004:12; Lag orders selected by AIC):
  - m-p: -1.9141
  - y: -0.1912
  - R: -2.1792
  - ∆p: -2.6661
  - ∆e: -2.6263
- Cointegrating tests (maximal eigenvalue and trace eigenvalue statistics with a degrees of freedom adjustment indicate rejection of no cointegration in favor of one cointegrating relationship):
  - Table 2 entries (as reported):
    - 0109.130.0056.860.00
    - 142.270.0530.520.02
    - 221.750.1317.150.17
    - 314.610.1112.340.10
    - 42.260.132.260.13
- Estimated cointegrating vector (Table 3; Dependent Variable: Real Broad Money; Sample period: 1998:1-2004:12; Frequency of data: Monthly; Number of lags used in the VAR: 6; R2: 0.82):
  - Real GDP: 0.4127*** (Std. Err. 0.106)
  - Time-deposit interest rate: -0.0007*** (Std. Err. 0.000)
  - Inflation: -1.2817*** (Std. Err. 0.326)
  - Depreciation: -0.1214* (Std. Err. 0.093)
  - Seasonal dummies: yes
  - Constant: yes
  - Significance notation: *, **, and *** correspond to the 10, 5 and 1 percent significance levels, respectively.
- Interpretation of coefficients (authors’ summary):
  - Coefficients of inflation and depreciation are negative and significant — goods and U.S. dollars are substitutes for money.
  - Estimated income elasticity ~ 0.4127, statistically not different from 0.5 and consistent with Baumol-Tobin transactions demand for money.
  - Time-deposit interest rate coefficient is very small, negative, and statistically significant at 1 percent; possible reasons: broad money composed primarily of interest-bearing deposits (most impact captured by inflation variable) and lack of variation in interest rate series.
- Diagnostics (Table 4 summary):
  - Tests performed: AR 1-5 test (portmanteau), Normality test (skewness and kurtosis), ARCH 1-5 test, Hetero test (heteroscedasticity).
  - Findings: estimated residuals free of autocorrelation and heteroscedasticity.
  - Normality cannot be rejected except for weak non-normality for real GDP and depreciation.
  - Recursive estimation (Figure 5) shows coefficients are constant over most of the sample.

### Out-of-sample/within-sample forecasts and structural break evidence
- Within-sample forecasts show that real money balances would not have been expected to increase by end-2004 as actually occurred (Figure 6).
- Residuals from the fitted money demand function reveal 2004 as an outlying year.
- Chow tests and structural break analysis:
  - Only one spike found in the test for structural changes in regression coefficients.
  - The structural break is in 2004 — the year when velocity deviated from the expected path.
  - Residuals suggest most observed points lie near a true money demand function, with 2004 departing significantly.

### Influence of exogenous factors on velocity movements and measurement issues
- Short-run influences on inflation and velocity noted:
  - Inertial component of inflation (lagged price response to monetary expansion).
  - Changes in demand for money (e.g., inflationary expectations) affecting velocity.
  - Wage policies, exogenous import price increases, and changes in administered prices affecting domestic relative prices.
  - Possible improper measurement of variables (CPI, money).
- Potential understatement of official inflation:
  - CPI may understate actual CPI because:
    - goods traded in the parallel market are not included in the CPI index;
    - CPI includes controlled and monitored prices;
    - CPI uses outdated weights from 1997.
  - Kovanen (2003) estimate: if full price liberalization had taken place in April 2003 with administered prices adjusted to market levels, overall CPI would have adjusted upward by about 47 percent.
- Additional factors cited:
  - Domestic energy prices were allowed to remain low despite increases in prices of energy imports.
  - Parallel market premium widened from 13 percent in January 2004 to about 53 percent in December 2004.
  - Pass-through to actual market prices may have been rapid anecdotally, but measured CPI may not have fully captured these increases.
- Net effect: these factors could have understated inflation during recorded disinflation in 2004 and help explain the dramatic fall in velocity.

### Conclusions and policy implications
- Main empirical conclusion:
  - Except for 2004, a stable demand for money as a function of parallel market exchange rate, inflation, and real output can be found in Zimbabwe.
  - Analysis shows a structural break in 2004: despite depreciation of the parallel exchange rate in 2004, real money balances increased.
- On explaining 2004 behavior:
  - A rapid fall in expected inflation could be one explanation, but this is atypical in disinflation episodes given inflation inertia and lagged adjustment of expectations.
  - Repressed inflation and mismeasurement of inflation are identified as the most likely possibilities.
  - The authors find no evidence of sustained disequilibrium behavior that would definitively confirm repressed inflation; final judgment requires a disequilibrium analysis.
- Policy guidance highlighted:
  - Important to coordinate monetary policy with management of the current price system.
  - In economies with price controls and distortions in the external sector, changes in money supply not accompanied by changes in domestic prices and exchange rate may produce repressed market demands and balance of payments disequilibrium.
  - Concern over magnitude of liquidity overhang built up in 2004 with M3 exceeding its reference.
  - Quoted policy lesson: "price controls do not get at the fundamental source of inflation—excessive expansion of the money supply—and eventually make the problem worse." Controls often result in shortages and bottled-up inflationary pressures that can explode.

*Source: Zimbabwean authorities and IMF staff estimates; IMF Working Paper content (1998:1-2004:12 sample; figures and tables as reported in the source).*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2006/_wp0615.pdf_