II. Income Velocity of Money in Ethiopia16
1. Understanding velocity is especially important when financial markets are underdeveloped and choices of monetary instruments are limited.17 With interest rates ineffective, monetary operations largely depend on targeting money aggregates. To set appropriate targets, a reasonable velocity path is vital. In Ethiopia, a working assumption in the past has been that velocity is declining by 2 percent each year, reflecting monetization and financial deepening (EDRI (2007). This allows the setting of money growth targets that are higher than the desired growth rates of nominal GDP.
2. Recent inflation in Ethiopia, however, raises serious doubts about the above approach.18 For 2004/05–2006/07, inflation consistently exceeded the levels implied by the differentials between broad money and real GDP growth. While broad money growth undershot levels that are consistent with single-digit inflation, assuming declining velocity, inflation averaged over 14 percent and was on average 7 percentage points higher than the differentials between broad money growth and real GDP growth (Table 1). This “puzzle” suggests a need to carefully examine the movements of the income velocity of money, particularly where inflation is high.
|1 Real GDP growth rate||12.6||11.6||11.4|
|2 Change in Velocity||-2.0||-1.9||-1.9|
|3 Inflation target||7.0||7.0||7.0|
|4 Implied money growth target||21.6||20.5||20.3|
|5 Money growth||19.6||17.4||19.7|
|6 Inflation (eop)||13.0||11.6||17.7|
|7 Change in velocity||2.73||5.35||8.41|
3. The quest for predictable movements in velocity is closely linked to studies on the determinants of and shifts in money demand. 19 Studies in developing countries generally find evidence for the stability of money demand, unit income elasticity, and a statistically significant impact of inflation on money demand. In the case of Ethiopia, Sterken (2004) examined transactions demand for narrow money (M1) for 1966–94. His findings confirmed that money demand was stable and inflation had a serious effect, but the estimated income elasticity was greater than one, which he attributed to unrecorded black market activities.
4. However, there is considerable skepticism about the validity of these studies of velocity/money demand in developing countries. Applying a similar approach in Ethiopia is particularly challenging because since independence Ethiopia has gone through a number of political and economic regime changes as well as numerous exogenous shocks (e.g., periodic droughts). Moreover, the lack of high-frequency data limits model specifications.
5. Recognizing the difficulties associated with time-series regressions, this paper adopts a less formal approach to studying velocity in Ethiopia. In particular, it relies on cross-country comparisons to put Ethiopia’s experience in a global context, and compare recent velocity movements in Ethiopia with trends in peer countries.
6. The paper is organized as follows. Section B examines how velocity evolves in the long run. Section C studies what affect velocity in low-income countries. Section D applies the discussion to Ethiopia. Section E concludes.
B. A Global View of Velocity Over Long Term
7. Bordo and Jonung (1987) argue that over time velocity should form a U-shaped curve (Figure 1), reflecting two opposite forces: monetization and financial innovation (emergence of money substitutes). Monetization—the rise of the monetary economy at the expense of barter—would increase demand for money. On the other hand, with modern financial innovations, complex transactions might require less money. The relative balance of these two forces would create velocity curves with three stages:
Figure 1.Income Velocity of Money
Stage 1: Velocity falls as monetization dominates.
Stage 2: Velocity is flat, with the two forces roughly balanced.
Stage 3: Velocity rises when financial innovation dominates.
8. Bordo and Jonung further argue that the velocity turning point for narrow money (M1) should come earlier than for broad money (BM) because of the switch from non-interest-bearing checking accounts to interest-bearing saving and time deposit accounts.
9. Following their methodology, velocity curves by income groups are constructed using data from 1995 to 2007 (Figure 2). Based on 2006 per capital GNI, IMF member countries/areas with complete GDP and money aggregates series are sorted into four groups: low-income countries (LICs); low-middle-income countries (LMICs); high-middle-income countries (HMICs); and high-income countries (HICs). Excluded from the sample are countries with hyperinflation, defined as 12–month inflation that exceeds 50 percent at any point.20 The results are broadly consistent with Bordo and Jonung’s findings for earlier data (1952–1982).
Figure 2.Broad Money Velocity by Income Groups (1995–2007)
BM velocity generally declines over time. During the sample period, 1995–2007, average BM velocity for every income group slopes downward. The cumulative declines range from 17 to 35 percent. Decline is broad based; only 18 of 137 countries in the sample had higher velocity at the end of the 13-year period.
BM Velocity declines more slowly as countries become more developed. This is consistent with the hypothesis that when a country reaches a more advanced stage, modern financial innovations and the increasing importance of nonbank financial institutions begin to counteract the impact on money demand and velocity of financial deepening.
Notwithstanding the common trends, the levels of BM velocity vary widely across countries (Table 2). Even within an income group, cross-country variations are very significant, probably because of circumstances that are unique to each country.
M1 velocity, although also sloping downward, is not strictly correlated with income (Figure 3). While the average is higher for HMICs than for LICs and LMICs, which supports the possibility of reversal of the velocity curve when a country becomes more advanced, the HICs actually have the lowest average M1 velocity of the four income groups. This probably reflects two reasons. One, the shift from checking to saving deposits may not be as significant as Bordo and Jonung have expected because more banks now pay interest on both. Second, since the introduction of Euro, M1 data are not available for Euro countries, which being at the high end of the development ladder, are more likely to see flat or even rising velocity.
|HI||27||1.4||3.2||0.3|Figure 3.M1 Velocity
10. For Ethiopia, BM velocity is relatively low compared with the LIC average (Figure 4). Before the recent rise, it was more in line with the LMIC average. While this level differential could be caused by many factors, one possibility is that nominal GDP is more under-recorded than in other LICs.
Figure 4.Broad Money Velocity
11. Rising BM velocity in Ethiopia since 2003 contrasts with the trend in other countries and is also a reversal to its own trend in earlier years. While most countries in all four income groups experienced a steady decline throughout the sample period, velocity in Ethiopia bottomed out in 2003 and has sloped upward since.
12. The low M1 velocity in Ethiopia relative to other countries is even more striking (Figure 3). Since compared with BM, M1 is more closely linked to transactions demand for money, it supports the idea that economic activities are significantly under-recorded. M1 velocity has been relatively stable, though showing a slight upward slope in recent years.
C. Which Variables Affect Velocity in Peer Countries?
13. Given that a country’s development level has a significant impact on the level and movements of BM velocity, the discussion of factors affecting velocity restricts its attention to LICs because their economic structure is more comparable to Ethiopia’s.
Opportunity cost of holding money (inflation)
14. While holding money facilitates transactions, it has costs—among them inflation, which reduces the purchasing power of money. It is thus to be expected that high inflation will reduce money demand and, in turn, lead to a higher velocity (money changes hands faster). This expectation is supported by data for LICs, which also seem to support the idea of adaptive behavior: people make decisions on the current money holding in part by inferring the current period inflation from the experience in the previous period. For example, a positive correlation between velocity and inflation is confirmed by a plot of yearly changes in velocity against inflation one year earlier for all 37 LICs for 1996–2007 (Figure 5) and on a more smoothed basis using average velocity change and inflation for the more recent period of 2003-07 (Figure 6).
Figure 5.Velocity Change and Lag Inflation (%) Figure 6.Average Velocity Change and Lag Inflation (%)
15. Simple regression analysis, based on the underlying data for Figure 5, suggests that a 1 percentage point increase in inflation increases velocity by 0.46 percent one year later. The implication is that, on average, when inflation exceeds about 14 percent, velocity is more likely to rise than fall. More generally, when inflation rates go up, the probability of rising velocity goes up even faster. (Table 3).
|Inflation Cutoff (%)||5||10||15||20||30|
|Prob. of rising velocity|
16. Velocity rising persistently for several years is rare. During the recent four years when velocity was rising in Ethiopia, only Haiti saw the same phenomenon. For 1995–2007, among all 37 LICs sampled, only five experienced rising velocity for four consecutive years, and only seven for three consecutive years (Table 4). Most of these countries had double-digit inflation during the period when velocity was rising, and, except for Kenya and Chad, all countries had far higher inflation when velocity was rising than the averages for the whole sample period. When velocity in these countries finally turned downward, most had made great strides in reducing inflation, often to single digits.
|Countries with Rising Velocity||Years||Period||Average||Maximum||Ave. (95-07)||Prior to Declining|
|Four consecutive years|
|Central African Republic||2000-03||1999-02||3.9||9.1||2.4||-1.3|
|Three consecutive years|
|Papua New Guinea||1998-00||1997-99||7.5||12.5||6.9||10.0|
Return on alternative investments
17. Existing literature has also studied velocity in the context of portfolio choice. Under this assumption, the opportunity cost of holding money can also be approximated by the return on holding alternative forms of wealth. While in financially more developed countries portfolio choices are quite wide, in most LICs, there are not many alternatives. T-bills are probably the most widely available investment vehicle with little credit risk. As expected, velocity increases with T-bill rates, which means lower money demand (Figure 7). The positive relation holds if nominal T-bill rates are replaced with real returns, defined as nominal rates minus CPI inflation (Figure 8). Ethiopia has the lowest nominal T-bill rate among all 18 LICs for which data are available. Together with double-digit inflation, this has driven the real return on T-bills in Ethiopia deep into negative territory—almost 6.5 percentage points below the next lowest one.
Figure 7.(2004–06 average) Figure 8.(2004-06 average)
Return on holding deposits
18. Velocity and money demand are also affected by the change in returns for holding money itself. Unlike M1, which mainly serves as the medium of exchange, BM includes a large amount of saving deposits, which compete with other types of investments as a store of value. Plotting average real deposit rates against the average velocity change for 2003–06 suggests that a higher real return on deposits causes velocity to decline and money demand to increase (Figure 9). In recent years, high inflation in Ethiopia has put a significant dent in the real return of saving deposits. Only two other LICs, Haiti and Guinea, had lower real returns, and both experienced double-digit inflation and rising velocity.
Figure 9.Real Deposit Rates and Change in Velocity (%)
D. Ethiopia’s case
increase in velocity in recent years (Table 5). Inflation in Ethiopia has generally been modest, spiking temporarily when harvests were poor. However, since 2004, inflation has largely stayed in doubt digits and continued to rise despite favorable harvests. Meanwhile, average inflation in LICs was low and continued to decline. The real interest rate on saving deposits in Ethiopia was about 8 percentage points below the LIC average.
|Inflation||Change in Inflation|
in percentage points)
|Impact on velocity||…||up ↑||up ↑||down ↓||up ↑|
20. The central role of inflation in explaining rising velocity in Ethiopia is more evident when the time series is longer (Figure 10 and 11). Before 2003, velocity in Ethiopia followed the LIC downward-sloping trend as its inflation rate was broadly below the LIC average. In 2003, inflation in Ethiopia exceeded the LIC average for the first time since 1999 and velocity began to move up, deviating from the LIC trend. The divergence continued after 2003 as inflation in Ethiopia stayed above the LIC average.
Figure 10.12–month Inflation Figure 11.Broad Money Velocity
21. The components of broad money reveal clues about why high inflation in recent years has led to a decline in money demand and rising velocity. With inflation persistently high, holding money for transactions became increasingly costly as its purchasing power eroded. Almost all demand deposits earn no interest in Ethiopia and daily transactions can be handled more easily by cash; high inflation would certainly favor cash over demand deposits. Indeed, within M1, there has been a small but steady increase in the share of currency in circulation (CIS), from 34 percent in 2002 to 36 percent in 2006 (Figure 12). The small magnitude of the change could be explained by the fact that Ethiopia was already a heavily cash-based economy.
Figure 12.Currency in Circulation
22. Inflation also eroded the credibility of using money as a store of value. Because the real interest rate on deposit accounts was significantly negative and declining (Figure 13), people have been switching to consumption or alternative investments to hold their wealth. Consequently, the disappearance of saving deposits has led to a noticeable rise of CIS as a share of BM (Figure 14). The ratio of saving and time deposits to nominal GDP has been declining since 2002 (Figure 15). With alternative financial investment in Ethiopia limited, high inflationary expectations appear to have caused a shift away from future consumption (savings) toward current consumption (Figure 16).
Figure 13.Real Return on Deposits (%) Figure 14.Currency in Circulation Figure 15.Saving and Time Deposits Figure 16.Private Consumption
23. The decline in money multiplier, which accompanied rising velocity, has reduced the effectiveness of monetary policy (Figure 17). As persistent high inflation pushed people away from bank deposits in recent years, the money multiplier reversed its rising trend and has headed downward. Thus, a given size intervention by the monetary authorities would have less impact on broad money growth and inflation.
Figure 17.Money Multiplier
24. Ethiopia’s shift to rapid growth and likely structural breaks in economic performance do not appear to account for the switch in velocity trend. In almost all years over the past 25 years, when velocity was rising, inflation was high or rising (Figure 18). The experience in other fast growing economies in Africa and Asia also confirm a robust correlation between inflation and velocity (Box 1).
Figure 18.Ethiopia: Velocity Change and Inflation (%) Box 1.Inflation and Velocity
Movements of velocity and inflation in China, one of fastest growing transition economies, are positively correlated. Over the last three decades, China saw five noticeable spikes of inflation: 1984–85, 1987–88, 1993-94, 2003–04, and 2006–07. Without exception, they were all accompanied by rising broad money velocity starting in the second year of the inflation surge.
The experience in Vietnam, another fast-growing Asian transition economy, although not as clear-cut as China’s, is also consistent with the view that inflation tends to reduce demand for money. When Vietnam experienced double-digit inflation in the mid-1990s, broad money velocity rose. Since then, velocity has turned downward. However, the size of yearly decline in velocity seems to be negatively correlated with inflation. When inflation is high, the size of decline is smaller and vice versa.
Successful disinflation in Tanzania, a strong growth performer in Africa, also put velocity back on a downward path. In the second half of the 1990s, Tanzania experienced a long period of double-digit inflation and strong growth. During this period, broad money velocity increased every year. Just before the turn of the century, inflation in Tanzania was brought under control and fell to single digits. One year later, velocity started to decline and has continued to do so as inflation remains relatively low. Growth remained strong after inflation was reduced.
25. Experience for Ethiopia and a broad range of countries point to high inflation being associated with rising velocity. Therefore, looking ahead, conservative money growth targets will be needed to offset the impact of rising velocity caused by Ethiopia’s high inflation. A monetary program based on a declining velocity or even a neutral assumption on velocity (no change) is likely to continue to lead to excess money supply and higher inflation than targeted. If the authorities are to achieve their goal of single-digit inflation, a much more conservative money growth taking into account the recent experience of rising velocity is called for.
26. Measures to move real interest rates to positive territory would also help restore the credibility of money as a store of value and create an environment for more effective monetary operations. Moreover, healthy growth of bank deposits will help promote financial intermediation and meaningful interest rates would help allocate the economy’s resources more effectively.
Bordo, Michael and LarsJonung (1987), “The Long-Run Behavior of the Velocity of Circulation: the International Evidence”, CambridgeUniversity Press.
Ethiopian Development Research Institute (EDRI) (2007), “Determinants of Inflation in Ethiopia.”
The main contributor to this chapter is Zaijin Zhan.
Income velocity of money in this paper is defined as the ratio of nominal GDP to money aggregates (broad money or narrow money).
Unless otherwise noted, all data for Ethiopia are for fiscal years from July 8 to July 7 (e.g., 2007=2006/07).
Money demand for a given income level is inversely related to velocity. In the short term, however, it is difficult to separate changes in velocity due to money supply and changes due to money demand. For example, a sudden increase in money supply would at first reduce velocity but over time higher inflation will reduce money demand and lead to an increase in velocity.
See appendix I for the countries used in calculation for each income group.