A. Introduction

1. A good understanding of the transmission mechanism of monetary policy in Sri Lanka has become more important since the adoption by the Central Bank of Sri Lanka (CBSL) of the goal to reduce inflation in the context of a flexible exchange rate regime. Moreover, with fiscal policy constrained by high public debt, monetary policy is the main countercyclical policy. This chapter explores how monetary policy affects other macroeconomic variables, mainly output and inflation. The first section provides an overview of the framework for implementing monetary policy while the following section discusses the transmission mechanism itself. Empirical estimates indicate that changes in policy interest rates have a significant effect on output but a small impact on inflation. Credit, however, does not respond strongly to changes in policy rates. The dominance of stateowned banks, weak capital bases in some banks, and the high level of non-performing loans (NPLs) constrain this channel of transmission.

B. Framework for Monetary Policy

2. The CBSL has increasingly focused on price stability. While attention continues to be placed on economic growth, the current balance represents a clear shift from past practices. In the 1980’s, emphasis was placed on developmental objectives—providing subsidized credit to priority sectors—and, until the introduction of the flexible exchange rate regime in January 2001, the CBSL aimed at preserving exchange rate stability. In line with this change in emphasis, the CBSL is considering whether to establish price stability as the single primary objective of the CBSL in the new Central Bank law that is expected to replace the Monetary Law Act (MLA) by end–2004.

3. The decision-making process of monetary policy has also become more transparent. Recent steps taken in this direction include the creation of the Monetary Policy Committee, which makes recommendations on which the Monetary Board acts, and the publication of the annual monetary program and monthly reviews analyzing its execution. Increased transparency, which has enhanced the credibility and predictability of the CBSL, is especially important while the CBSL remains not fully independent from the government (the Secretary of the Treasury is a member of the Monetary Board and the MLA allows the Minister of Finance to direct the CBSL to adopt certain policies).

4. To achieve its objectives, the CBSL monetary program targets the rate of growth of reserve money. In particular, the CBSL has aimed, indirectly, at a rate of expansion of broad money in line with expected growth in nominal GDP. As the money multiplier has remained broadly unchanged in recent years (after accounting for changes in reserve requirements), this has translated into a similar target for reserve money growth.

5. Open market operations are the CBSL’s main tool to maintain reserve money on its targeted path. These consist of (i) standing overnight repo and reverse repo facilities (repo rates constitute the main signal of the stance of monetary policy); (ii) since March 2003, daily auctions of repo or reverse repos at amounts determined by the CBSL; and (iii) outright sales and purchases of government or CBSL securities. Other instruments at the CBSL’s disposal to influence monetary conditions are statutory reserve requirements on commercial bank deposit liabilities, foreign exchange operations, and moral suasion. Reserve requirements, however, are not used in practice for active liquidity management—they were gradually lowered until they reached 10 percent on rupee deposits in 2001 and have remained unchanged since then—and, currently, forex interventions are restricted to building up the stock of reserves and to limiting excessive volatility.

6. Overall, the institutional framework for monetary policy in Sri Lanka seems appropriate for the pursuit of low inflation. Reserve money functions as a nominal anchor that the CBSL can effectively target with the instruments available. Recent steps to increase the transparency of monetary policy decisions have strengthened the credibility and predictability of the CBSL. Further steps in this direction could include establishing price stability as the single primary objective in the new Central Bank law and strengthening the independence of the CBSL.

C. Transmission Mechanism

7. The CBSL relies mostly on the interest rate and credit channels for the transmission of monetary policy to prices and output. Thus, a drop in policy rates is expected to lead to a decline in market-determined interest rates and to an increase in lending that would in turn have a positive effect on economic activity while putting upward pressure on inflation. Alternatively, one can put this story in terms of money: if the CBSL engineers an (unexpected) expansion of the money supply, activity and prices will tend to rise. The following sections explore to what extent such a mechanism is operational in Sri Lanka.

The Broad Picture

8. The table below contains some descriptive statistics for the variables of interest for the observation period (Q1-1995 to Q2-2003).

• On average, quarter-on-quarter, real GDP expanded by 1 percent, inflation was about 2.3 percent, and broad money increased by 3.4 percent—entirely consistent with the quantity theory and a very small decline in the velocity of money. (These growth rates have not been annualized.)

• Inflation has been extremely volatile—the 95 percent confidence interval for the unconditional expectation of (quarterly) inflation extends from -0.7 percent to 5.2 percent.

• Reserve money has grown on average by less than broad money as reserve requirements have declined gradually.

• The exchange rate depreciated on average by 2 percent a quarter, which was consistent with a broadly constant real effective exchange rate.

• Overall credit expanded faster and was more volatile than credit to the private sector, on account of occasional accelerations in credit to government and public corporations.

Descriptive Statistics of Selected Variables, Q1-1995 to Q2-2003

Sources: Sri Lankan authorities; and Fund staff estimates. Note: Δ denotes the quarter-on-quarter difference in the logarithm of the variable (seasonally adjusted). The sample covers Q1-1995 to Q2-2003 (quarterly GDP for 1995 was interpolated).

Descriptive Statistics of Selected Variables, Q1-1995 to Q2-2003

	Mean	Median	Maximum	Minimum	Standard Deviation
Repo rate	11.9	11.5	19.3	8.5	2.8
3-month t-bill rate	13.4	12.5	19.5	8.6	3.0
12-month t-bill rate	14.0	13.3	20.5	8.6	3.0
Prime rate	15.9	15.1	22.5	10.9	2.8
Δ Colombo CPI	2.3	2.3	6.5	-1.3	1.8
ΔGDP	1.0	1.1	2.4	-2.1	1.0
Δ reserve money	2.0	2.1	6.0	-4.0	2.0
Δ broad money	3.4	3.3	6.3	0.3	1.2
Δ overall credit	3.6	3.5	9.0	-1.4	2.2
Δ private credit	3.1	2.9	8.1	-2.4	1.8

Sources: Sri Lankan authorities; and Fund staff estimates. Note: Δ denotes the quarter-on-quarter difference in the logarithm of the variable (seasonally adjusted). The sample covers Q1-1995 to Q2-2003 (quarterly GDP for 1995 was interpolated).

View Table

Descriptive Statistics of Selected Variables, Q1-1995 to Q2-2003

	Mean	Median	Maximum	Minimum	Standard Deviation
Repo rate	11.9	11.5	19.3	8.5	2.8
3-month t-bill rate	13.4	12.5	19.5	8.6	3.0
12-month t-bill rate	14.0	13.3	20.5	8.6	3.0
Prime rate	15.9	15.1	22.5	10.9	2.8
Δ Colombo CPI	2.3	2.3	6.5	-1.3	1.8
ΔGDP	1.0	1.1	2.4	-2.1	1.0
Δ reserve money	2.0	2.1	6.0	-4.0	2.0
Δ broad money	3.4	3.3	6.3	0.3	1.2
Δ overall credit	3.6	3.5	9.0	-1.4	2.2
Δ private credit	3.1	2.9	8.1	-2.4	1.8

Sources: Sri Lankan authorities; and Fund staff estimates. Note: Δ denotes the quarter-on-quarter difference in the logarithm of the variable (seasonally adjusted). The sample covers Q1-1995 to Q2-2003 (quarterly GDP for 1995 was interpolated).

9. Changes in policy rates appear to translate into changes in market-determined interest rates. Simple regressions (ordinary least squares) provide evidence that changes are transmitted quickly throughout all maturities, with shorter maturity rates responding the fastest.² One cannot reject the hypothesis that the coefficient estimates in each of the regressions add up to one. In the case of the 3–month T-bill rate, one cannot even reject the hypothesis of full pass through within the same quarter.

$\begin{array}{ccc}\mathit{\Delta }\mathit{tbill}3\hfill & =& \underset{\left(0.09\right)}{0.89}\mathrm{\Delta }\mathit{repo}+\underset{\left(0.09\right)}{0.10}\mathit{\Delta }\mathit{repo}\left(-1\right)\hfill \\ \mathit{\Delta }\mathit{tbill}13\hfill & =& \underset{\left(0.09\right)}{0.73}\mathrm{\Delta }\mathit{repo}+\underset{\left(0.09\right)}{0.14}\mathit{\Delta }\mathit{repo}\left(-1\right)\hfill \\ \mathit{\Delta }\mathit{prime}\hfill & =& \underset{\left(0.08\right)}{0.66}\mathrm{\Delta }\mathit{repo}+\underset{\left(0.08\right)}{0.22}\mathit{\Delta }\mathit{repo}\left(-1\right)\hfill \end{array}$

10. Simple correlations of interest rates and money with lags and leads of output and prices also have the expected sign, consistent with a transmission mechanism at work.

• Three to five quarters after a rise in repo rates inflation tends to decline. Contemporaneously, however, inflation is higher, which is consistent with either causality from high inflation leading the CBSL to raise rates or with the so called “price puzzle” (a stylized fact that some have explained because of a short term increase in the cost of working capital caused by higher interest rates).

• A rise in repo rates is associated with lower output in the following year, with the strongest correlation two to three quarters after the policy change.

• Changes to repo rates are negatively correlated with the supply of money, with a rise in rates leading the fall in the money supply by one quarter.

Cross-Correlations of Changes in the Repo Rate with Selected Variables

Sources: Sri Lankan authorities; and Fund staff estimates. Note: Δ denotes the quarter-on-quarter difference in the logarithm of the variable (seasonally adjusted). The sample covers Q1-1995 to Q2-2003 (quarterly GDP for 1995 was interpolated).

Cross-Correlations of Changes in the Repo Rate with Selected Variables

	t-4	t-3	t-2	t-1	t	t+1	t+2	t+3	t+4
Δ CCPI	-0.31	-0.38	-0.02	0.15	0.22	0.08	-0.08	-0.01	-0.11
Δ GDP	0.17	0.27	0.29	0.20	-0.07	-0.27	-0.53	-0.43	-0.12
Δ broad money	-0.10	-0.04	0.09	0.01	-0.12	-0.29	-0.18	-0.12	0.06

Sources: Sri Lankan authorities; and Fund staff estimates. Note: Δ denotes the quarter-on-quarter difference in the logarithm of the variable (seasonally adjusted). The sample covers Q1-1995 to Q2-2003 (quarterly GDP for 1995 was interpolated).

View Table

Cross-Correlations of Changes in the Repo Rate with Selected Variables

	t-4	t-3	t-2	t-1	t	t+1	t+2	t+3	t+4
Δ CCPI	-0.31	-0.38	-0.02	0.15	0.22	0.08	-0.08	-0.01	-0.11
Δ GDP	0.17	0.27	0.29	0.20	-0.07	-0.27	-0.53	-0.43	-0.12
Δ broad money	-0.10	-0.04	0.09	0.01	-0.12	-0.29	-0.18	-0.12	0.06

Sources: Sri Lankan authorities; and Fund staff estimates. Note: Δ denotes the quarter-on-quarter difference in the logarithm of the variable (seasonally adjusted). The sample covers Q1-1995 to Q2-2003 (quarterly GDP for 1995 was interpolated).

• When the money supply is expanding at a relatively high pace, inflation tends to rise in the following two quarters.

• An expansion in the money supply appears also related with higher output four and five quarters thereafter.

Cross-Correlations of Changes in Broad Money with Selected Variables

Sources: Sri Lankan authorities; and Fund staff estimates. Note: Δ denotes the quarter-on-quarter difference in the logarithm of the variable (seasonally adjusted). The sample covers Q1-1995 to Q2-2003 (quarterly GDP for 1995 was interpolated).

Cross-Correlations of Changes in Broad Money with Selected Variables

	t-4	t-3	t-2	t-1	t	t+1	t+2	t+3	t+4
Δ CCPI	0.13	0.08	-0.20	-0.20	0.02	0.31	0.53	0.14	-0.20
Δ GDP	-0.12	-0.16	-0.06	0.23	-0.04	-0.07	-0.17	0.02	0.21

Sources: Sri Lankan authorities; and Fund staff estimates. Note: Δ denotes the quarter-on-quarter difference in the logarithm of the variable (seasonally adjusted). The sample covers Q1-1995 to Q2-2003 (quarterly GDP for 1995 was interpolated).

View Table

Cross-Correlations of Changes in Broad Money with Selected Variables

	t-4	t-3	t-2	t-1	t	t+1	t+2	t+3	t+4
Δ CCPI	0.13	0.08	-0.20	-0.20	0.02	0.31	0.53	0.14	-0.20
Δ GDP	-0.12	-0.16	-0.06	0.23	-0.04	-0.07	-0.17	0.02	0.21

Sources: Sri Lankan authorities; and Fund staff estimates. Note: Δ denotes the quarter-on-quarter difference in the logarithm of the variable (seasonally adjusted). The sample covers Q1-1995 to Q2-2003 (quarterly GDP for 1995 was interpolated).

More Details: A VAR with Output, Prices, Money and Interest Rates

11. To better explore the monetary transmission mechanism, a vector autoregression model (VAR) was estimated. The regression includes four lags of GDP, prices, broad money, the repo rate and a constant term. The VAR impulses are orthogonalized by using the standard Choleski decomposition by imposing the following ordering: GDP is assumed not to react to contemporary shocks to other variables; the price level responds to past shocks of all variables but only contemporary shocks to output; broad money comes next; and finally repo rates, which react to contemporary shocks of all variables.³ The results are illustrated through impulse response functions (Figure II.1)

12. An unexpected increase in the repo rate by 100 basis points (equivalent to one standard error of the innovations) is followed by a further 50 basis points increase in rates the following quarter, which suggests that changes in the policy stance come about somewhat gradually. Eventually, about a year after the initial innovation, the policy move is partially reversed as the central bank reacts to the impact of its actions on money, output, and inflation.

13. The tightening of the monetary stance slows the growth rate of broad money gradually—by about ⅓ percent a year after the rise in repo rates by 100 basis points. The impact on economic activity is relatively large and fast: GDP declines by ½ percent in the second quarter after the innovation (this is relative to the level it would have been in the absence of a monetary policy shock) and by about 1 percent one year after. That is, if the economy was cruising at a growth rate of 5 percent a year and repo rates rise unexpectedly by 100 basis points, one would expect GDP growth to decline to about 4 percent one year after the event. The results of the variance decomposition for GDP indicate that monetary shocks account for less than 20 percent of its variability at any frequency.

14. The effect on prices, however, is largely muted as inflation declines by less than ½ percent even six quarters after the monetary tightening. The large weight of food prices in the price index is responsible for this result. Replacing GDP by agricultural GDP in the VAR above confirms that most of the contraction in GDP that follows a monetary tightening is due to a contraction in nonagricultural GDP. As a result, food prices do not change substantially as supply is not affected and demand is relatively inelastic. Since the weight of food in the price index is about ⅔, it is not surprising that the link between monetary policy shocks and inflation is weak. Consistent with this result, the variance decomposition for inflation indicates that shocks to output and inflation (roughly in equal proportion) explain most of its variability, rather than monetary shocks.

View Full Size

Accumulated Response to an Increase in the Repo Rate

(One Cholesky standard deviation)

Citation: IMF Staff Country Reports 2004, 069; 10.5089/9781451823493.002.A001

• Download Figure
• Download figure as PowerPoint slide

15. Two more results merit attention (Figure II.2). First, a positive shock to GDP translates into a large decline in inflation (the effect is smaller, however, if the shock is confined to nonagricultural GDP). Second, evidence of substantial inflation inertia—a shock to inflation generates a substantial further increase in prices for more than two years. Wage formation in the government sector, which in the past has been strongly backward looking, coupled with a central bank that has been willing to accommodate high inflation expectations are a probable cause for inflation inertia. This points toward the need to strengthen some institutions: wage negotiations, especially in the public sector, need to be forward looking while the central bank needs to make clear that it will be less willing to validate inflationary expectations (for instance, by credibly committing to make price stability its sole primary objective).

View Full Size

Accumulated Responses of Inflation

Citation: IMF Staff Country Reports 2004, 069; 10.5089/9781451823493.002.A001

• Download Figure
• Download figure as PowerPoint slide

Exploring the Credit Channel

16. To explore the role of the credit channel in the transmission of monetary policy, a VAR model including GDP, prices, bank credit to the private sector, the prime lending rate, and the repo rate is estimated (using four lags and the Choleski decomposition with this ordering of variables).

17. Private sector credit is estimated to decline only modestly after an innovation in the repo rate (Figure II.3). Thus, an initial increase in the repo rate of about 100 basis points, which eventually brings about a further cumulative increase in the repo rate of about 150 basis points, decreases private sector credit by about VV percent during the first year. Two years after the rise in the repo rate, private sector credit has declined by about ½ percent. Since the impact on GDP is relatively larger and faster—GDP is estimated to decline by almost 1 percent five quarters after the rise in the repo rate—the estimates suggest a weak role for the credit channel in the transmission of monetary policy. Prime rates do respond to changes in the repo rate but change only by about half as much.

View Full Size

Accumulated Responses to an Increase in the Repo Rate

(One Cholesky standard deviation)

Citation: IMF Staff Country Reports 2004, 069; 10.5089/9781451823493.002.A001

• Download Figure
• Download figure as PowerPoint slide

18. The dominance of state-owned banks in the banking system, low capital bases, and the high level of nonperforming loans (NPLs) constrain the monetary transmission mechanism. First, state-owned banks (which currently provide about one third of total credit to the private sector) are less likely than private banks to base their lending decisions on strictly commercial criteria and, more generally, have weaker incentives to behave in a profit maximizing fashion—which weakens the link between interest rates and the provision of credit. Second, a high level of nonperforming loans may influence interest rate and credit decisions. For instance, as the repo rate declines banks with weak balance sheets may resist or be slow to revise down their lending rates and extend new credit—in this way, they improve their profitability while not risking higher NPLs. This is more likely to occur when the NPL problem is widespread, which it is in Sri Lanka—affecting mostly state-owned banks but also some private banks. In addition, the limited supply of “bankable” lending opportunities can also make private credit fairly inelastic to the interest rate.

19. The weak recovery of private sector credit by state-owned banks since the recession of 2001 also underscores their need for restructuring. Overall credit to the private sector increased by almost 18 percent from December of 2001 to June of 2003, but most of the increase was accounted for by the operations of private banks—state-owned banks increased their lending by less than 10 percent during the same period. In addition to the reasons outlined above, this was also due to the inability of state-owned banks lacking risk management skills to adjust to a situation of receding demand for credit by the public sector.

20. The lack of depth and maturity of the nonbank sector may also impede the smooth transition of monetary policy impulses. In Sri Lanka, the nonbank sector is dominated by the public pension funds (primarily, EPF), which invest almost exclusively in government securities. A diversification of EPF’s investments away from treasury securities would contribute to deepening capital markets. This might also entail outsourcing portfolio management to competing private managers.

An Exploration of the Exchange Rate Channel

21. To explore the role of the exchange rate in the transmission of monetary policy, a VAR model including GDP, prices, the exchange rate, and the repo rate is estimated (using again the Choleski decomposition with this ordering of variables).⁴ The results are generally inconclusive but point to a very weak role for the exchange rate, which was not market determined for most of the sample period, in the transmission of monetary policy.

22. The estimated impulse responses produce the counterintuitive result that a rise in the repo rate is followed by a depreciation of the exchange rate—but this appears to be influenced by developments since late 2000 (Figure II.4). In 2000 and 2001, faced with a rapid expansion in public sector credit and pressures on the balance of payments, the CBSL tightened interest rates and allowed a faster depreciation of the exchange rate—including by introducing a floating regime. Subsequently, steps toward fiscal consolidation allowed the CBSL to lower interest rates while the rate of depreciation of the exchange rate declined. Before 2000, the exchange rate exhibited a steady depreciation with low variability from quarter to quarter. When the VAR is re-estimated for the sub-sample from 1995:1 to 2000:2, a tightening of interest rates still appears associated with an exchange rate depreciation but the effect is quantitatively trivial (perhaps also in part because of the reduced sample size). These results can also be explained by the relative closed nature of the financial account.

23. An unexpected depreciation (of less than 2 percent), in turn, is associated with a significant decline in output (by more than 1½ percent), which suggests that such an innovation may be a proxy for some underlying macroeconomic weakness. As in the previous section, interest rate innovations are estimated to affect output—in this case, a rise in repo rates by 100 basis points reduces both by more than ½ percent one year later. The pass-through of the exchange rate to prices is also significant (about two-thirds after three quarters).

24. Of course, if loosening monetary policy is not estimated to lead to a depreciation of the exchange rate, lower policy interest rates can not stimulate the export sector through a depreciation of the currency—i.e., the exchange rate channel has not been a significant one for the transmission of monetary policy. In the future, however, given the importance of the export sector in the Sri Lankan economy, the gradual liberalization of the financial account and with a floating exchange rate and macroeconomic stability, the exchange rate channel is bound to gain importance in the transmission of monetary policy.

View Full Size

Accumulated Responses to an Increase in the Repo Rate

(One Cholesky standard deviation)

Citation: IMF Staff Country Reports 2004, 069; 10.5089/9781451823493.002.A001

• Download Figure
• Download figure as PowerPoint slide

D. Conclusions

25. Empirical estimates indicate that changes in policy interest rates have a significant effect on output but a small impact on inflation. In particular, an unexpected increase in the repo rate by 100 basis points lowers the growth rate of GDP by about ½ percentage point after six months and by about 1 percentage point after a year. Inflation in Sri Lanka is very volatile and the effect of monetary shocks is dwarfed by supply-side shocks, in particular to agricultural production given the large weight of food in the CPI. There is evidence of substantial inflation inertia, caused in part by backward looking wage settlements in the public sector (which influence wage formation in the private sector). In the future, the focus of wage negotiations should be placed on productivity rather than on past inflation.

26. The credit channel plays a weak role in the transmission of monetary policy. Credit does not respond strongly to changes in policy interest rates—despite market interest rates closely following changes in policy rates. The dominance of state-owned banks, weak capital positions in some banks, and the high level of NPLs constrain this channel of transmission. To strengthen it, it is necessary to:

• Move decisively with the restructuring of state-owned banks, in the case of People’s Bank urgently given its precarious position. Privatization remains the best option to place relations between the government and the state banks at arm’s length. The government is planning to divest People’s Bank. Bank of Ceylon, which is solvent, is meeting most of its restructuring targets, including on loan recovery and return on assets.

• Strengthen the capital base of both state and private banks. The CBSL should take prompt action if a bank falls short of capital requirements.

• Address the high level of NPLs, which requires banks to write-off long-standing delinquent loans, increase their provisions, and step up their risk management capabilities. At the same time, debt recovery laws need to be strengthened and recent constitutional doubts over the bank’s powers for parate execution (i.e., their ability to recover collateral without having to obtain a court judgment) need to be resolved. In this area, the AMC law may facilitate recovery from NPLs.

• More generally, increase the efficiency of financial intermediation. The banking sector is likely to witness increased consolidation in the near future and, as a result, increase its efficiency. The new Asset Management Company Law will help banks restructure their balance sheets, while amendments to the banking laws (which were tabled in Parliament in November) apart from enhancing the CBSL’s supervisory powers, will also facilitate the merger of banks.

A. Introduction

27. Workers’ remittances and tourism play important roles in the Sri Lankan economy.⁶ This chapter investigates the economic effects of remittances and tourism, and the government’s role in these areas. Workers’ remittances and tourism receipts have averaged 7 percent of GDP per annum over the last two decades, with remittances particularly important as a steady and growing source of much needed foreign exchange (especially when contrasted with other, more volatile sources). While the two-decade long civil conflict adversely affected tourism, looking ahead this sector appears to be poised for strong expansion, particularly if lasting peace is achieved. The government places great value on the economic contribution of these sectors, and has thus implemented various policies to encourage their development. Remittances, which so far have primarily supported consumption, could enhance Sri Lanka’s domestic growth potential if used to finance investment. This would be more likely if the environment for business, particularly for small-scale entrepreneurs, was made friendlier. Tourism has significant potential for growth if the peace process succeeds. Given the sector’s low capacity utilization, it is well placed to accommodate growing demand. The development of tourism in the longer-run would require significant investment, particularly to alleviate infrastructure and transportation bottlenecks.

View Full Size

Foreign Inflows

(In percent of GDP)

Citation: IMF Staff Country Reports 2004, 069; 10.5089/9781451823493.002.A001

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Volatility of Inflows

(coefficient of variation)

Citation: IMF Staff Country Reports 2004, 069; 10.5089/9781451823493.002.A001

• Download Figure
• Download figure as PowerPoint slide

B. Workers’ Remittances

28. Data on remittances over the last two decades reveal the following features:

• Remittances have averaged around 5½ percent of GDP per annum, and an annual growth rate of around 10 percent. Since 1982, remittances have totaled $11½ billion, compared to $10¾ billion and $2 billion in donor assistance and FDI, respectively. Importantly, remittances have been substantially less volatile than other inflows.

• This has made remittances an important source of financing for the balance of payments. Remittances were on average sufficient to finance almost half of trade deficits, or more than 85 percent of debt service payments.

• Remittances are highly correlated with private consumption. Investment funding out of remittances is largely limited to housing and education.

• Remittances have been a major factor in poverty alleviation. Roughly 15 percent of the total population benefit from remittances.⁷ About half of all emigrants come from poor areas.⁸ The income-augmenting role of remittances has been particularly important in the North and East.⁹ The government’s PRSP, “Regaining Sri Lanka,” also stresses the poverty alleviation role played by income transfers of migrant workers. Taken together with the close relationship between consumption and remittances discussed earlier, this suggests that the poor are particularly vulnerable to external developments that reduce remittances.

View Full Size

Remittances and Consumption

(Changes in percent)

Citation: IMF Staff Country Reports 2004, 069; 10.5089/9781451823493.002.A001

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Regional Migration and Poverty

(In percent of total)

Citation: IMF Staff Country Reports 2004, 069; 10.5089/9781451823493.002.A001

• Download Figure
• Download figure as PowerPoint slide

Some of the determinants of remittances have been:

• In terms of demand for Sri Lankan workers, economic conditions in Middle Eastern oil-producing countries are a major factor. The majority of Sri Lankan emigrants have gone to the Middle East, with emigration to Saudi Arabia, Kuwait, and the United Arab Emirates accounting for 95 percent of total outflows in recent years. Emigration to Asia and Europe is small with the corresponding outflows of 3½ percent and 2 percent of total emigration. In terms of the flow of remittances, the share from the Middle East is only 60 percent of the total, compared with just over 20 percent from Europe, which reflects the higher average earnings of emigrants to Europe (most likely due to their higher skill level). Thus, one of the major factors affecting the flow of migrant earnings has been oil price volatility.¹⁰

  

  

  View Full Size

  

  Emigration and Remittances, 2001

  (In percent of total)

  Citation: IMF Staff Country Reports 2004, 069; 10.5089/9781451823493.002.A001

  • Download Figure
  • Download figure as PowerPoint slide

  

  

  View Full Size

  

  Composition of Emigrants by Skill Level

  (In percent of total emigration)

  Citation: IMF Staff Country Reports 2004, 069; 10.5089/9781451823493.002.A001

  • Download Figure
  • Download figure as PowerPoint slide

• On the supply side, the lack of employment opportunities at home has motivated Sri Lankans to seek jobs overseas. The civil conflict, the demographic structure, and low economic growth in rural areas resulted in relatively high levels of unemployment. Reports suggest that the earnings of temporary unskilled workers in the Middle East could be 8 times higher than wages of similarly qualified workers in Sri Lanka.¹¹

• The temporary nature of Sri Lanka’s emigration and the skill and gender composition of emigrants ensure a steady flow of remittances. The academic literature in this area suggests that the propensity to remit is higher when emigration is temporary, and when the worker’s skill level is lower. Most Sri Lankans work on 2-3 year contracts with a possibility of renewal, returning home after the expiration of the contract. Almost three-quarters of emigration are of low-skilled workers, with housemaids making up more than half of the total. Professionals and skilled workers represent much smaller shares, of ½ percent and 20 percent of annual emigration, respectively. Women comprise more than two-thirds of Sri Lankan emigrants. The majority of women emigrants are from rural areas, a reversal from the 1980s when emigration was predominantly an urban phenomenon. The majority of migrants are people under 40 years of age with families. Thus, the welfare of the household is one of the most important factors affecting Sri Lankans’ decision to migrate.¹²

• The growth of remittances has also been aided by regulatory and policy changes. Regulations on travel and foreign exchange convertibility were relaxed in the late 1970s. Transfers through the banking system—covering an estimated 85 percent of all remittances—have become less expensive and more convenient over time. In this regard, the possibility of keeping remittances in domestic or foreign currency has been helpful. The Sri Lanka Bureau of Foreign Employment (SLBFE) has been providing migrants with consulate services, training and insurance coverage.¹³ The government has prepared a national employment strategy (May 2002) that focuses on overseas employment and envisages policy and institutional changes to (i) upgrade migrant worker skills; (ii) revamp training programs to meet foreign employment requirements; and (iii) provide information, as well as legal and financial assistance to potential and returning migrants. The strategy envisages setting up a network for monitoring employment opportunities abroad.¹⁴

• In addition, foreign employment opportunities have increased through improved access to foreign labor markets. In addition to visa-free travel to Middle Eastern countries, Malaysia recently agreed to provide at least 50,000 jobs to Sri Lankans beginning in 2004, while Italy and Korea have extended employment quotas for 800 and 1,000 workers from Sri Lanka, respectively. Negotiations are ongoing on visa-free travel within the region. Sri Lanka has bilateral agreements with Jordan, Lebanon, and Qatar regulating the rights of Sri Lankan workers in these countries.

29. While the welfare benefits of remittances have been substantial, foreign employment and earnings of migrant workers are susceptible to exogenous shocks, especially because the majority of overseas workers are unskilled. The share of skilled workers in Sri Lanka’s emigrants has been rising only gradually. In addition, Sri Lankan migrants face stiff competition from similar workers from Indonesia, India, Pakistan, Bangladesh, and the Philippines. However, even temporary foreign employment can help migrant workers to obtain and upgrade skills. Moreover, such workers often set up small businesses upon their return to Sri Lanka, thus creating jobs. In this regard, improvement in the business environment and facilities for small businesses are important areas where government policies can help to maximize the domestic economic potential of workers who have returned to Sri Lanka.

View Full Size

Tourist Arrivals

(In thousands)

Citation: IMF Staff Country Reports 2004, 069; 10.5089/9781451823493.002.A001

• Download Figure
• Download figure as PowerPoint slide

C. Tourism

30. Tourism has been adversely affected by the civil conflict, as security concerns have deterred tourists from visiting Sri Lanka. This in turn has depressed investment in the sector. Tourist arrivals had been growing at an average rate of 22 percent during 1975-1982, but the trend was reversed in 1983 when the conflict broke out. It then took 16 years to return to the levels of early 1980s. The attack on the Colombo airport in July 2001 further dampened tourism, with arrivals dropping by 15 percent that year. A sharp decline in tourism generally coincided with intensification of the conflict. During the conflict period, tourism contributed only modestly to foreign exchange flows, and receipts were quite volatile.

31. However, since the ceasefire agreement in early 2002, the sector’s performance has improved markedly. Tourist arrivals rose by 17 percent in 2002 and tourism receipts reached 2 percent of GDP (against 1⅓ percent of GDP the previous year). The trend continued in 2003 as tourist arrivals rose by 25 percent in the first six months, compared to a year earlier. However, even with the recovery, hotel occupancy—at 43 percent in 2002—was still below its 1998 level of 58 percent. More recent reports suggest that this situation has improved considerably.

32. Tourism has a significant upside potential in Sri Lanka. The tourism industry aims to increase arrivals to 1 million by 2010. However, several hurdles need to be overcome:

• On the supply side, lasting peace is clearly the most important factor for this sector’s expansion as evidenced by the strong recovery in the last two years.

• Investment in infrastructure and hotels is needed. Poor roads have been a major constraint on the sector’s expansion, but expressways to major tourist destinations are currently projected for completion by 2007, thus easing transportation bottlenecks. Other requirements for the sector’s expansion include adequate and affordable supply of electricity, water supply, and solid waste removal. The authorities have identified priority areas for tourism development, primarily in the South and West, and are working with donors to increase their financing of infrastructure rehabilitation in these areas.

• The expansion of tourism requires an increase in the number of flights to Sri Lanka. The country’s “open-sky” regime is expected to attract more carriers.

• Tourism remains highly dependent on an improved global economic outlook, especially in the EU. Tourists from the EU account for almost 50 percent of total arrivals. Economic conditions in India, which accounts for 20 percent of tourists to Sri Lanka, are also important.

• Sri Lanka needs to improve its international image as a tourist destination. The Sri Lanka Tourist Board is taking a number of initiatives in this regard. These include the expansion of the visa-free entry regime (now available to citizens of 73 countries), and efforts to diversify tourism away from budget travelers, and toward cultural, nature, sports, religious, and eco-tourism.

A. Pros and Cons of Common Tax Incentives

34. Tax incentives can be granted in a variety of ways, including tax holidays, investment tax credits, and accelerated depreciation.¹⁶ Accelerated depreciation is generally considered the best in terms of minimizing revenue losses, distortions in investment decisions, and tax abuse, while tax holidays are considered the worst in this regard.

• Corporate income tax (CIT) holidays fully exempt qualified projects from the CIT. They can be quite costly: they entail a deadweight tax loss (as the investment might have occurred without the incentive), and tax avoidance may result from transfer pricing. They also result in reduced transparency and accountability, as tax-exempt companies are rarely required to file proper tax returns. However, they are popular because they reduce tax administration, allow investors to avoid onerous tax regulations and corrupt tax officials, and are neutral in their impact on the capital/labor mix of qualified projects. Preferential CIT rates are a form of partial tax holiday, with lower revenue costs and greater transparency as investors remain in the tax net.

• Investment tax allowances/credits are cost-recovery incentives aimed primarily at investment in plant and equipment in targeted industries. They are generally better-targeted instruments than CIT holidays, and their revenue costs are more transparent and easier to control. Investment allowances typically stipulate that a certain percentage of the initial cost of the investment can be deducted as an expense in the current period in addition to contributing to depreciation allowances in future periods. Investment credits allow a share of the investment to be deducted from tax liabilities. However, they tend to distort investment in favor of short-lived assets, since a further credit or allowance becomes available when the asset is replaced. The incentives can also be abused by qualified enterprises that claim multiple allowances through the sale and purchase of the same asset, or who act as a purchasing agent for unqualified enterprises.

• •Accelerated depreciation entails lower revenue costs and fewer incentives for abuse. Since accelerating the depreciation of an asset does not affect the total depreciation of the asset, little distortion in favor of short-lived assets is generated. There is also little incentive for enterprises to engage in the kind of tax abuse connected with investment credits/allowances.

B. Main Features of Sri Lanka’s Tax Incentives Regime

35. As part of its export promotion strategy, Sri Lanka introduced a tax incentives regime in the late 1970s. This regime was administered through the Board of Investment (BOI).¹⁷ BOI-registered companies were typically located within export processing zones (EPZs), and were exempt from certain provisions of the Inland Revenue Act, Customs Ordinance, and exchange control regulations. In addition, these companies had the right to employ expatriates, to borrow abroad, and to repatriate capital from their investments. In 1982, the BOI also took over the customs functions within the zones, allowing BOI-registered companies to circumvent the perceived inefficiencies of the Customs Department.

36. At the beginning of the 1990s, tax incentives offered under the BOI Act were extended well beyond the export-oriented sector, to offer a wide range of incentives aimed at general investment promotion.¹⁸ These currently include tax holidays, concessionary corporate income tax rates, generous depreciation allowances, and import duty and excise exemptions.

37. In addition to incentives provided under the BOI Act, Sri Lanka also offers tax incentives under the Inland Revenue Act. While originally these incentives were focused on areas not covered by the BOI Act, there has been significant overlap in recent years. Tax incentives under the Inland Revenue Act cover the following areas: the manufacture and export of goods deemed nontraditional exports; agriculture, fisheries and agro-processing (other than tea and prawn culture); industrial tools and machinery; small- and large-scale infrastructure projects; the rehabilitation of under-performing industries; and companies engaged in R&D. The incentives usually take the form of tax holidays of 2-12 years’ duration, depending on the sector, and a period of lower corporate tax rates once the tax holidays have expired. Sri Lanka also offers significant accelerated depreciation allowances, and import duty exemptions on raw materials and capital goods.

38. Some modest restrictions on Sri Lanka’s tax incentives regime have been imposed recently. In late 2002, the BOI was brought under the auspices of the Inland Revenue Act, thereby eliminating its ability to modify provisions of the Act.¹⁹ In addition, the government announced in late 2003 that the extension of certain tax holidays beyond their expiration date will be curtailed beyond 2004, and all registered companies, including those registered with the BOI, will be required to file tax returns from 2004 onwards.

C. Tax Incentives and FDI in a Regional Context

36. Data limitations constrain a rigorous analysis of the actual effect of tax incentives on attracting FDI to Sri Lanka. In the absence of detailed data, a brief assessment of the relationship between tax incentives and FDI flows is undertaken for a selected group of countries in Asia.

40. The use of tax incentives, especially to attract FDI, remains widespread in Asia.²⁰ Special economic zones (SEZs) and EPZs are quite common, as are tax holidays, although there is considerable variation in their length and coverage. More detailed information on the incentives regimes of a selected group of Asian countries is given below.

• India offers a wide range of exemptions and incentives. Export-oriented undertakings qualify for tax holidays of up to 10 years provided they are located in an economic zone/park. Tax incentives are also granted for infrastructure projects, for developing industrial parks and special economic zones, and for the generation and transmission of power. Industrial undertakings in under-developed states, telecommunication services, and scientific R&D companies qualify for tax holidays of up to five years.

• Indonesia offers tax holidays of 3-8 years for new enterprises in specific sectors. Incentives are awarded by a ministerial committee with nontransparent criteria. Capital goods are both duty and VAT exempt. Indonesia also has numerous SEZs.

• Malaysia offers narrowly available tax holidays and widely available allowances for investments in manufacturing, agriculture, hotels and tourism, research and development, and training activities. Companies granted pioneer status can receive an income tax holiday of 70 percent. Certain other companies, such as those in the high tech sector, can still qualify for significant investment and accelerated depreciation allowances. Malaysia has also established numerous SEZs.

• Thailand has increased the number and generosity of tax incentives since the Asian financial crisis. For the most part, Thailand’s incentives policy is geared towards regional development. Tax holidays range from 3 years for export industries in the most developed regions, to 8 years plus a 50 percent reduction in corporate tax for the following five years, for companies that locate in the least developed region of the country. An investment allowance of 25 percent for infrastructure spending is also available. Incentives are awarded largely on a discretionary basis following an assessment by the Thai Board of Investment. Thailand also has numerous EPZs.

41. Countries that offer a larger set of incentives, in particular tax holidays, have not enjoyed higher FDI. India and Sri Lanka, which offer the widest-ranging tax holidays, have significantly lower FDI than Malaysia, which offers a narrow range of tax holidays combined with significant investment allowances

Comparison of FDI Performance in Sri Lanka with Selected Countries, 1996-2000

Source: UNCTAD (2003).

Comparison of FDI Performance in Sri Lanka with Selected Countries, 1996-2000

	FDI Inflows Per Capita (in U.S. dollar)	FDI Inflows (as a percent of gross fixed capital formation)
India	2.7	2.9
Bangladesh	1.2	1.7
Malaysia	227.1	16.9
Pakistan	4.5	6.5
Sri Lanka	12.4	5.9
Thailand	56.9	11.0
Vietnam	23.4	25.1

Source: UNCTAD (2003).

View Table

Comparison of FDI Performance in Sri Lanka with Selected Countries, 1996-2000

	FDI Inflows Per Capita (in U.S. dollar)	FDI Inflows (as a percent of gross fixed capital formation)
India	2.7	2.9
Bangladesh	1.2	1.7
Malaysia	227.1	16.9
Pakistan	4.5	6.5
Sri Lanka	12.4	5.9
Thailand	56.9	11.0
Vietnam	23.4	25.1

Source: UNCTAD (2003).

42. Countries with a weaker macroeconomic and socio-political environment seem to offer more generous tax incentives. As noted by Chalk (2002),²¹ the Philippines and Indonesia, whose business climate is less attractive compared to most of its regional competitors, offer the most generous. The same seems to apply for Sri Lanka, with its severe labor market rigidities, a tax administration perceived to be inefficient and corrupt, and its long civil conflict. Indeed, the generosity of Sri Lanka’s tax incentives regime increased since the mid-1990s, when the civil conflict intensified.²²

43. The effectiveness of tax incentives in Asian countries may depend less on the characteristics of the incentives themselves and more on the characteristics of the countries in which they are used. In countries with strong civil service institutions, the detrimental aspects of tax incentives may be more limited than in other countries. An OECD report (1994) concurs with this view, concluding that although tax incentives were important promotional tools in some Asian countries, they did not play an overriding role in promoting FDI.²³ Investors tended to place greater value on the fundamental conditions for their investments such as infrastructure, general business environment, human resources, the legal system, and other supporting services.

D. Tax Incentives and Corporate Income Tax Performance

Comparison of Corporate Tax Rates, Revenues and Efficiency Ratios, 2001 ^1/

Sources: Price Waterhouse Coopers (2001); and Fund staff estimates.

^1/Defined as revenue to GDP divided by the standard tax rate, expressed in percent. In the case of India, the domestic rate is used and for Indonesia the top CIT rate is taken.

Comparison of Corporate Tax Rates, Revenues and Efficiency Ratios, 2001 ^1/

	Rate	Revenue to GDP	Efficiency Ratio
India	35	1.6	4.6
Indonesia	10,15,30	1.6	5.3
Malaysia	28	6.2	22.1
Phillipines	32	2.7	8.4
Singapore	25.5	6.2	24.3
Sri Lanka	35	1.3	3.7
Thailand	30	2.9	9.7
Vietnam	32	5.6	17.5

Sources: Price Waterhouse Coopers (2001); and Fund staff estimates.

^1/Defined as revenue to GDP divided by the standard tax rate, expressed in percent. In the case of India, the domestic rate is used and for Indonesia the top CIT rate is taken.

View Table

Comparison of Corporate Tax Rates, Revenues and Efficiency Ratios, 2001 ^1/

	Rate	Revenue to GDP	Efficiency Ratio
India	35	1.6	4.6
Indonesia	10,15,30	1.6	5.3
Malaysia	28	6.2	22.1
Phillipines	32	2.7	8.4
Singapore	25.5	6.2	24.3
Sri Lanka	35	1.3	3.7
Thailand	30	2.9	9.7
Vietnam	32	5.6	17.5

Sources: Price Waterhouse Coopers (2001); and Fund staff estimates.

^1/Defined as revenue to GDP divided by the standard tax rate, expressed in percent. In the case of India, the domestic rate is used and for Indonesia the top CIT rate is taken.

44. Following the expansion of its tax incentives regime during the 1990s, Sri Lanka’s corporate tax performance has become one of the worst in the region, in terms of both overall CIT revenue and the efficiency of the corporate tax.²⁴ Other countries that offer wide-ranging tax holidays have also exhibited weak CIT performance. India and Indonesia performed poorly, with Singapore and Malaysia performing the best. This in part reflects the type of tax incentives these countries offer. As noted in Zee et al. (2002), wide-ranging tax holidays, compared to investment cost-recovery incentives such as accelerated depreciation allowances, have the greatest detrimental effect on revenue, due to deadweight losses and high levels of tax abuse.²⁵

45. Rationalizing corporate tax incentives could entail considerable revenue gains. The authorities’ tax expenditure study suggests that the direct cost of corporate income tax concessions amounts to 0.4 percent of GDP.²⁶ About 75 percent of this cost arises from companies enjoying a zero corporate tax rate, mostly under the auspices of the BOI. Revenue losses from customs duty exemptions granted to BOI companies are estimated at around of 0.5 percent of GDP. The total cost (including tax evasion) is likely to be significantly higher.

Revenue Loss from Tax Incentives

(In percent of GDP; unless otherwise indicated)

Source: Authorities’ Tax Expenditure Study, November 2003; and staff estimates.

Revenue Loss from Tax Incentives

(In percent of GDP; unless otherwise indicated)

Corporate income tax concessions	0.4
Preferential rates	0.1
Zero rates	0.3
Customs duty exemptions	0.5

Source: Authorities’ Tax Expenditure Study, November 2003; and staff estimates.

View Table

Revenue Loss from Tax Incentives

(In percent of GDP; unless otherwise indicated)

Corporate income tax concessions	0.4
Preferential rates	0.1
Zero rates	0.3
Customs duty exemptions	0.5

Source: Authorities’ Tax Expenditure Study, November 2003; and staff estimates.

E. Concluding Remarks

46. Given the considerable erosion in revenue that has occurred in recent years, Sri Lanka can ill afford its current tax incentives regime. Moreover, now that a ceasefire is in place, one of the principal justifications previously given for offering tax incentives—to compensate for a poor political and business environment—is less valid. Indeed, FDI has rebounded since the ceasefire, with flows in the first half of 2003 increasing to $170 million, compared with $55 million during the same period in 2002. Looking ahead, the following recommendations can be made:

• Tax incentives, especially tax holidays, should be rationalized.²⁷ Given the weakness of Sri Lanka’s institutions, the cost of the tax incentives—including lower revenue and distorted investment decisions—are far greater than the possible benefit from higher FDI. Since depreciation allowances are already high, the government should consider compensating for the reduction in these incentives by introducing a single low CIT rate.

• To improve tax administration, an effective revenue authority that streamlines and improves customs clearance and tax collections should be set up. This will help sustain a policy of rationalizing tax incentives. (One of the key reasons why the policy efforts to reduce tax incentives were not sustained during the 1990s was the perception that customs administration was weak and that companies wanted to bypass it.)