International Monetary Fund Washington, D.C.

Transcription

1 2009 International Monetary Fund June 2009 IMF Country Report No. 09/192 Kenya: Seleted Issues and Statistial Appendix This Seleted Issues paper and Statistial Appendix for Kenya was prepared by a staff team of the International Monetary Fund as bakround doumentation for the periodi onsultation with the member ountry. It is based on the information available at the time it was ompleted on Deember The views expressed in this doument are those of the staff team and do not neessarily reflet the views of the overnment of Kenya or the Exeutive Board of the IMF. The poliy of publiation of staff reports and other douments by the IMF allows for the deletion of market-sensitive information. Copies of this report are available to the publi from International Monetary Fund Publiation Servies th Street N.W. Washinton D.C Telephone: (202) Telefax: (202) publiations@imf.or Internet: International Monetary Fund Washinton D.C.

2

3 INTERNATIONAL MONETARY FUND KENYA Seleted Issues and Statistial Appendix Prepared by Mr. Kalina (head) Mr. MIntyre Mr. Chen (all AFR) Ms. Aylward (PDR) Robert Thaidze and Ms. Lusinya (FAD) Approved by Afrian Department Deember Contents Pae I. Growth Performane in Kenya Durin A. Introdution...5 B. Stylized Fats about Kenya s Growth Performane...5 C. A Growth Aountin Exerise for Kenya...7 D. Determinants of Year-on-Year TFP Growth durin E. Poliy Impliations...14 Appendix: Sensitivity Analysis...15 II. Prie Dynamis in Kenya Durin A. Introdution...16 B. Bakround...16 C. Key Determinants of Inflation...18 D. Eonometri Analysis...18 E. Poliy Impliations...21 III. Estimation of the Equilibrium Real Exhane Rate for Kenya...23 A. Introdution...23 B. Model...25 C. Results...28 D. Poliy Impliations...30 IV. The Wae Bill and Civil Servie in Kenya...32 A. Introdution...32 B. Size of the Wae Bill...32 C. Sustainability of the Current Wae Bill...35 D. Wae and Publi Setor Employment Struture...37 E. Conlusions...39

4 2 V. Trade Interation in the East Afrian Community...40 A. Introdution...40 B. Trade Flows and Trade Reimes in the EAC...40 C. The EAC Customs Union...45 D. Trade Impat: An Assessment for Kenya...50 E. Other Reasons for East Afrian Interation...55 F. Conlusions...57 Appendix: The Smart Simulation Model...58 Referenes...63 Text Fiures I.1. Comparative Growth Performane I.2. Produtivity I.3. TFP Governane and Inflation I.4. TFP Growth and Maroeonomi Indiators II.1. Inflation II.2. Potential Fators Underpinnin Inflation II.3. Impat of the Reression Variables on Inflation III.1. Exhane Rates and Relative CPI III.2. Eonomi Fundamentals Underpinnin the Equilibrium REER III.3. Atual and Equilibrium REER V.1. Reional Interation Arranements in Afria...39 Text Tables I.1. Setorial Contributions to Real GDP Growth I.2. Comparative Growth Performane I.3. Estimates of Real GDP and Fator Inputs I.4. Results of Growth Aountin Exerise I.5. OLS Estimates of a Redued-Form Reression I.6. Growth Aountin Exerise: Sensitivity Analysis II.1. A Comparison on Inflation with Other Countries II.2. Contributions to Overall Inflation by Components II.3. OLS Estimates of a Redued-Form Inflation Equation III.1. Equilibrium (Cointeratin) Relation between the REER And the Eonomi Fundamentals IV.1. Cross-Reional Comparisons of Central Government Waes and Salaries IV.2. Publi Setor Wae Bill Relative to Other Maroeonomi Indiators...32 IV.3. Publi Servie Wae Bills in Seleted Sub-Saharan Afrian Countries...33 IV.4. Simulations Under a Stati Current Poliies Senario...34 IV.5. Employment in the Publi Setor IV.6. Estimated Real Averae Earnins in KSh per annum...36 IV.7. Compression of the Wae Struture...37

5 3 V.1. EAC Countries: Exports and Imports V.2. EAC Countries: Reional Trade by Commodities V.3. Features of Trade Reimes of Kenya Tanzania and Uanda...44 V.4. EAC Countries: Evolution of Tariff Reimes V.5. EAC Countries: Estimated Effets of Proposed Tariff Chanes...51 V.6. Trade Simulation Results...51 Statistial Appendix Tables 1. Gross Domesti Produt by Oriin at Constant Pries Gross Domesti Produt by Oriin at Current Pries Expenditure on Gross Domesti Produt at Constant Pries Expenditure on Gross Domesti Produt at Current Pries Gross Domesti Produt GDP Deflator Population and Real Per Capita GDP Gross Fixed Capital Formation at Current Pries Sales of Ariultural Prodution to the Marketin Boards Value of Ariultural Prodution Sold to the Marketin Boards Averae Pries to Produers for Seleted Commodities Quantity Index of Manufaturin Output Seleted Statistis on Constrution Ativity Enery Supply-and-Demand Balanes Employment by Industry and Setor Averae Wae Earnins per Employee by Industry and Setor Employment and Earnins in the Publi Setor Consumer Prie Index Deember Central Government Fisal Operations 1997/ / Central Government Revenue 1999/ / Eonomi Classifiation of Central Government Expenditure and Net Lendin 1999/ / Funtional Classifiation of Central Government Expenditure and Net Lendin 1997/ / Loal Government Finanes 1997/ / Gross Domesti Debt of the Central Government 1996/ / Operatin Profits and Cash Position of Seleted Publi Enterprises 1998/ / Central Bank of Kenya Balane Sheet Deember 1999 September Monetary Survey Deember 1999 September Commerial Banks Liquidity June 1999 September Nonbank Finanial Institutions Liquidity June 1999 September Prinipal Interest Rates Marh 2001 September Distribution of Credit to Private Setor June Balane of Payments Tea Prodution and Exports Coffee Prodution Consumption and Exports

6 4 33. Commodity Composition of Trade Trade Volumes and Pries Value Unit Value and Volume of Major Exports Destination of Exports Commodity Composition of Imports Imports by Country of Oriin External Servies Inome and Transfer Aounts External Debt Indiators Appendix I. Tax Summary as of Deember

7 5 I. GROWTH PERFORMANCE IN KENYA DURING A. Introdution 1. This hapter examines Kenya s rowth performane durin Speifially it onsiders the followin issues: first the stylized fats about Kenya s rowth performane in the past two deades in omparison with other Sub-Saharan Afrian ountries; seond the main soures of eonomi rowth in Kenya in the ontext of a onventional rowth aountin exerise; third the main determinants of the results indiated in the rowth aountin exerise; and finally the key poliy impliations. 2. The followin stylized fats emered from reent studies on the rowth performane in Sub-Saharan Afria: 2 The rowth performane of the reion has been weak in the past few deades; The main soure of eonomi rowth in the reion has been fator aumulation with rowth in TFP playin little role; and TFP rowth tends to be positively orrelated with hih quality institutions ood overnane and sound maroeonomi poliies. 3. The main findins of this hapter are the followin: Sine the early 1990s Kenya s eonomi performane has been weaker than the averae for Sub-Saharan Afrian ountries and the weakest amon the three members of the East Afrian Community (EAC); 3 As in other Sub-Saharan Afrian ountries Kenya s rowth has been driven mostly by fator aumulation with total fator produtivity (TFP) delinin markedly in the past two deades; and The low TFP rowth over the past two deades has been sinifiantly assoiated with poor overnane and hih inflation. B. Stylized Fats about Kenya s Growth Performane 4. Kenya s eonomi performane has been lakluster. Durin real GDP rowth averaed around one perent per annum. Growth was robust durin the 1980s when real GDP rowth averaed 4.5 perent per annum but delined notably in the 1990s averain 1.9 perent. The main soure of rowth durin the 1980s and 1990s was the tertiary setor. (Table I.1). 5. Kenya s rowth performane has slipped behind its neihbors sine the 1990s. As indiated in Fiure I.1 and Table I.2 Kenya s eonomi performane was well above its 1 This hapter was prepared by Kevin C. Chen (AFR). 2 For details see Tahari Ghura Akitoby and Aka (2004). 3 The three members of the EAC are Kenya Tanzania and Uanda.

8 6 East Afrian neihbors and the averae for all developin ountries in the 1980s. However sine the 1990s Kenya s rowth performane has been weaker than most other developin ountries and underperformed both Tanzania and Uanda. Durin the ap between Kenya s rowth performane and its East Afrian neihbors has widened further. Table I.1. Kenya: Setorial Contributions to Real GDP Growth / 1980s 1999s (In perent of total GDP rowth) Primary setor Seondary setor Tertiary setor / Contribution is alulated by the share of the setor in total ouput multiplied by the rwoth rate of the setor. Soure: Kenyan Authorities Fiure I.1. Kenya: Comparative Growth Performane (1980=100) Least Developed Countries Kenya Sub-Sahara Exludin Nieria & South Afria

9 7 Table I.2 Kenya: Comparative Growth Performane / Averae Annual Growth Number of Years of Growth (In Perent) Number of Years of Deline Hiher than 4 perent 1980s 1990s Past Ten Years Past Ten Years Kenya Tanzania Uanda Developin Countries Sub-Sahara Afria 2/ Least Developed Countries World / Kenyan Data were provided by the Kenyan Authorities. Data for other ountries were provided by the World Eonomi Outlook (WEO) database. Areate roups were defined by WEO database. 2/ Exludin Nieria & South Afria Methodoloy and Data C. A Growth Aountin Exerise for Kenya 4 6. The rowth aountin exerise deomposes the real GDP rowth into the rowth of total fator produtivity (TFP) and fator aumulation inludin rowth in physial apital human apital as well as total employment. 5 Followin most studies a Cobb-Doulas prodution funtion is assumed for the Kenyan eonomy. Speifially α 1 α t ( Lt H t ) Y = A K (1) t t where Y is ross domesti produt in real terms A is the total fator produtivity (TFP) K is the physial apital stok L is total employment and H is an index of human apital stok. 4 The soure of eonomi rowth matters beause if the main soure of rowth is fator aumulation then aordin to the law of diminishin returns in fator inputs lon-term rowth is not sustainable. For details see Kruman (1994) and Youn (1995). 5 A rowth aountin exerise was implemented for Kenya durin in the ross-ountry study of Tahari Ghura Akitoby and Aka (2004). This hapter adds the followin to the existin literature: first it examines more losely the movements of TFP rowth in Kenya durin ; seond it separates the rowth of human apital from the TFP and treats it as a fator input; finally it examines key fators sinifiantly assoiated with TFP rowth in Kenya durin the past two deades.

10 8 The parameter α is the inome share of apital whih is assumed to be Takin loarithms and differentiatin we obtain the followin rowth aountin equation: ΔY Y = ΔA ΔK ΔL ΔH + α + ( 1 α) (1 α) (2) A K L H Equation (2) deomposes the rowth rate of output into the rowth rates of TFP physial apital total employment and human apital. 7. Data for output physial apital labor total employment and human apital are displayed in Table I.3. Physial apital K is alulated by the onventional perpetual inventory method as disussed in Barro and Sala-i-Martin (2000): t+ 1 = I t + (1 δ K t (3) K ) where I is the level of real investment and δ is the rate of depreiation of the existin apital stok. Given estimates of the depreiation rate and the initial apital stok as well as a time series for real investment the apital stok series is alulated reursively usin (3). In this study the depreiation rate is assumed to be 6 perent whih is well within the rane of 4 10 perent used in similar studies. The ratio of apital to GDP is assumed to be 2 in The human apital index is alulated as follows: H t = w jt s jt (4) j where s jt is the proportion of workers with eduation level j where j varies from 0 (orrespondin to no shoolin) to 6 (orrespondin to ompletion of tertiary eduation). is the relative wae orrespondin to workers with eduation level j. Data on eduation attainments were obtained from Barro and Lee (2001) and relative wae orrespondin to different eduation levels were alulated based on data on the return to shoolin found in Appleton Bisten and Manda (1999). w jt 6 Senhadji (2000) found that the inome share of physial apital in the Sub-Saharan Afria reion was around The sensitivity analysis in the Appendix shows that relaxin this assumption does not sinifiantly alter the main results. 7 These assumptions will be relaxed in the sensitivity analysis whih indiates that the rowth aountin exerise is robust to these assumptions.

11 9 Table I.3. Kenya: Estimates of Real GDP and Fator Inputs Year GDP Investment Capita1 Employment 1/ Human Capital In thousands of In billions of Kenya shillin onstant 1982 pries persons Index Proj Soure: Central Bureau of Statistis of Kenya 1/ Inludes only the formal setor. Results 8. Kenya s fator produtivity durin has been disappointin. As indiated in Fiure I.2 apital produtivity defined as GDP/K delined durin refletin investment ineffiieny. While labor produtivity defined as GDP/L exhibited an upward trend in the 1980s it was sluish durin the 1990s. 9. Like most Sub-Saharan Afrian ountries Kenya s eonomi rowth appears to have been primarily driven by fator aumulation. As indiated in Table I.4 whih summarizes the estimates derived from equation (2) the deline in total fator produtivity appears to have aounted for the sluish rowth of the Kenyan eonomy refletin effiieny losses typial of eonomies plaued by strutural weaknesses. An important issue is therefore the identifiation of the key fators that have ontributed to the deline in Kenya s total fator produtivity.

12 Fiure I.2. Kenya: Produtivity (1990=100) Index Capital Produtivity Labor Produtivity Year Table I.4. Kenya: Results of Growth Aountin Exerise Annual Averae Annual Averae Contributions to Output Growth Growth Rate of Phsysial Total Human Period Output Capital Employment Capital TFP of whih: Proj Soure: Staff estimates

13 11 D. Determinants of Year-on-Year TFP Growth durin Potential fators affetin the year-on-year TFP rowth in Kenya durin the past two deades inlude: 9 Governane Durin the last two deades Kenya has been plaued by pervasive problems of internal onflits onstitutional rises and orruption sandals. All of these are likely to have undermined the rowth of TFP. 10 Maroeonomi environment The positive link between a favorable maroeonomi poliy environment and hih eonomi rowth is well doumented in the rowth literature Observations of Fiure I.3 suest that movements of TFP rowth have been sinifiantly orrelated with overnane and inflation. 12 Speifially TFP rowth has been positively assoiated with ood overnane but neatively with inflation. 13 For example TFP rowth was larely neative in the early 1990s amidst immense politial instability and hih inflation. 14 On the other hand the robust TFP rowth in 1995 was assoiated with ood overnane and low inflation. In addition followin the eletion in late 2002 of President Kibaki who promised zero tolerane on orruption TFP rowth bean to rise in Other maroeonomi variables however do not bear suh obvious and strikin relations with TFP rowth durin (Fiure I.4) The sample period was trunated in 1984 beause some key variables suh as overnane were not available prior to While HIV/AIDS is a key fator affetin Kenya s lon-term rowth performane it is unlikely to affet the year-on-year movements of TFP rowth. 10 The positive link between TFP and sound institutions as well as ood overnane is well doumented in the rowth literature. See for example Bosworth and Collins (2003) and Rodrik Subramanian and Trebbi (2002). 11 See for example Senhadji (2000). 12 Governane is alulated as the averae of three politial risk indiators ompiled by The International Country Risk Guide (ICRG): orruption law and order as well as internal onflit. The hiher the index the better the performane. 13 While inflation is an important variable on its own it an also be interpreted as a proxy of the overall soundness of maroeonomi poliy stane beause hyperinflation as ourred in the early 1990s also tends to reflet a poor maroeonomi environment at lare. 14 The early 1990s was a period of deep politial fissures. For example in 1990 the forein minister was murdered and riots broke out in the summer. Also the 1992 eletions ended up in reat soial turmoil. 15 Given the small sample size the result should not be interpreted as suestin that other fators are not important to the TFP rowth in Kenya. In fat other fators may well be the dominant fators determinin TFP rowth durin another sample period.

14 12 Fiure I.3. Kenya: TFP Governane and Inflation TFP (Annual perent hane) Governane (Index a hiher index orresponds to better overnane) Inflation (In perent) Soure: Staff estimates.

15 13 Fiure I.4. Kenya: TFP Growth and Maroeonomi Indiators TFP (Annual perent hane) Terms of Trade (1995=100) Reserve Assets (In months of imports) Current Aount Balane (In perent of GDP) Soure: Staff estimates Fisal Balane (In perent of GDP) Openness (Ratio of exports and imports to GDP In perent)

16 A simple eonometri model supports the above observations on the relations between TFP and overnane as well as other maroeonomi variables. 16 The model is estimated by ordinary least squares usin annual data for A eneral-tospeifi priniple is utilized in the reression analysis: initially a eneral model enompassin all variables that may potentially affet TFP overnane inflation openness to trade fisal indiators as well as external indiators was estimated. Thereafter variables found to be statistially insinifiant were eliminated sequentially. 13. The reression results suests that overnane and inflation appear to have been sinifiantly orrelated with TFP (Table I.5). Table I.5. Kenya: OLS Estimates of a Redued-Form Reression TFP Growth Reression Explanatory Variables Constant Governane Inflation Coeffiient Estimates Absolute t-statistis (3.70) (3.44) (3.38) Number of Observations: 21 R 2 : 0.49 Prob(F-Statistis) 0.00 Soure: IMF staff alulations. Note: Sample period is The dependent variable is the annual rowth rate of TFP. The explanatory variables are Governane and Inflation. Governane is measured by the averae of three indiators onsistin of orruption law and order and internal onflit ompiled by The International Country Risk Guide. Inflation is the 12-month perent hane of the CPI. The fiures in parentheses are absolute t-statistis based on standard errors alulated usin Newey- West heterosedastiity and autoorrelation onsistent ovarianes. E. Poliy Impliations 14. The above results lend support to the overnment s onoin efforts to strenthen overnane. The overnane aenda fouses on several reforms inludin upradin the publi budet and finanial manaement systems strenthenin the anti-orruption institutions and improvin the judiial framework. Moreover the overnment s onoin efforts to reform the politial system is interal to the overall overnane aenda. 15. Maintainin prie stability should be the overridin objetive of the monetary poliy. Considerable aution therefore needs to be exerised when usin monetary poliy for ounter-ylial purposes. 16 Results should be interpreted with aution in liht of the small sample size of the reression analysis. Also owin to the small sample size a simple ordinary least squares estimation was used here instead of a fuller VAR model typially used in studies on ross-ountry differenes in TFP.

17 15 Appendix Sensitivity Analysis 16. Althouh the above rowth aountin exerise is based on arbitrary assumptions about the initial apital stok and its rate of depreiation sensitivity analysis suests that the results are robust to different assumptions. Table I.6.A shows the results for the rowth aountin exerise under a different assumption about the initial apital stok with a apital/gdp ratio in 1963 of one as opposed to a ratio of two assumed in the benhmark senario. Likewise hanes in the depreiation rate (Table I.6.B) or inome share of apital (Table I.6.C) do not alter the main results. Table I.6. Kenya: Growth Aountin Exerise: Sensitivity Analysis (In Perent) Annual Averae Annual Averae Contributions to Output Growth Growth Rate of Phsysial Total Human Period Output Capital Employment Capital TFP A. Assumin real apital to GDP ratio equals one in B. Assumin a depreiation rate of 4 perent C. Assumin apital share of inome equals 60 perent Soure: Staff estimates

18 16 II. PRICE DYNAMICS IN KENYA DURING A. Introdution 1. Kenya s inflation has reently risen sharply. Annual overall inflation as measured by the twelve-month perent hane in the onsumer prie index (CPI) rose from 8 perent in September 2003 to 19 perent in September Underlyin inflation (overall inflation exludin food and enery pries) reahed 7 perent from 3 perent a year earlier exeedin the Central Bank of Kenya (CBK) s taret of 5 perent. 2. This hapter examines key determinants of Kenya s prie dynamis durin the past deade. A simple eonometri model suests that key determinants of inflation durin have inluded broad money rowth food rop output movements of the nominal effetive exhane rate (NEER) and the international ommodity pries of fuel and enery. Aainst this bakround the reent episode of hih inflation appears to have been triered by the exessively loose monetary onditions a poor harvest hih enery pries and a weakenin Kenyan shillin. B. Bakround 3. The reent episode of hih inflation is the most severe sine After peakin at 61 perent in early 1994 inflation delined substantially in 1995 to around 1.6 perent (Fiure II.1). Between Kenya s inflation averaed around 7 perent per annum and has been lower than the averaes for Sub-Saharan Afria as well as other developin ountries (Table II.1). However inflation has reently aelerated substantially reahin a histori hih sine 1995 in September. 4. Durin the past deade flutuations in food pries aounted for the bulk of the movements in the CPI (Table II.2). With food and nonaloholi beveraes arryin more than 50 perent of the weiht in the CPI basket food supply onditions play a sinifiant role in Kenya s prie developments. In this onnetion more than perent of the inreases in the CPI durin reent months have been attributed to rises in food pries. Underlyin inflation whih exludes food has also risen substantially in reent months. 2 1 This hapter was prepared by Kevin C. Chen (AFR). 2 Three onepts of underlyin inflation are urrently used for Kenya. The Central Bureau of Statistis of Kenya ompiles a measure that exludes only food and nonaloholi beveraes (FNB). In addition to exludin FNB the underlyin inflation used by the Central Bank of Kenya to uide monetary poliy exludes fuel and power (FP) as well as transport and ommuniation (TC). The underlyin inflation presented in the Fund s Staff Reports exludes FNB as well as FP while inludin TC.

19 Fiure II.1.Kenya: Inflation (12-month perent hane) OVERALL 10 UNDERLYING 5 0 Mar-95 De-95 Sep-96 Jun-97 Mar-98 De-98 Sep-99 Jun-00 Mar-01 De-01 Sep-02 Jun-03 Mar-04-5 Table II.1. Kenya: A Comparison on Inflation with Other Countries averae averae Kenya Tanzania Uanda Sub-Sahara Afria Least Developed Countries Developin Countries World Table II.2. Kenya: Contributions to Overall Inflation by Components Jan Feb Mar Apr May Jun Jul Au Sep Food & Nonaloholi Beveraes Alohol & Tobao Clothin & Footwear Housin Fuel & Power Household Goods & Servies Medial Goods & Servies Transport & Communiation Rereation & Eduation Personal Goods & Servies Total

20 18 C. Key Determinants of Inflation Intuitive Reasonin 5. Both demand and supply fators appear to have ontributed to the reent spike in inflation (Fiure II.2). These fators inlude: Low food rop output The drouht experiened in parts of the ountry durin May-September is estimated to have taken a heavy toll on food rop output. The prodution of maize prodution the most important ommodity in the diet of the Kenyans is estimated to have dereased by 14 perent in Hih enery pries While fuel and enery diretly aounts for less than 5 perent in the CPI basket the effets of the sure in world enery pries ould have been onsiderable as the indiret effets suh as the inreased ost of transportin food have been sinifiant. Weakenin Kenyan Shillin The Kenya shillin has depreiated in nominal effetive terms by around 10 perent in the year to September 2004 thereby puttin upward pressure on import pries and inflation. Exessively loose monetary onditions The loosenin of monetary poliy sine July 2003 has resulted in sharply neative real interest rates and a 15 perent expansion in broad money (M3X) in the twelve months preedin September D. Eonometri Analysis 6. A simple eonometri model has been devised to apture the potential impat of these fators on prie dynamis in Kenya. Speifially the followin sinle redued-form equation was estimated usin quarterly data for the period : 5 inf=f(laed inf laed maize laed enery laed neer laed money) 3 Apart from its sinifiane in the Kenyan diet maize prodution is also a ood proxy for rop prodution enerally beause adverse weather with a sinifiant impat on maize prodution usually also affets the output of other food rops suh as wheat and beans. 4 In this paper the rowth rate of M3X in terms of the urrent exhane rate is used. The Staff Report presents money rowth rates in terms of a onstant proram exhane rate. 5 The sample is trunated in 1995 beause prior data over a period of extremely hih inflation peakin at over 60 perent in early Extendin the data to an earlier period ould potentially distort the estimates for the later period whih is the fous of this paper.

21 19 Fiure II.2. Kenya: Potential Fators Underpinnin Inflation Gross Prodution of Maize (12-month perent hane) 120 World Enery Commodity Pries (12-month perent hane) Mar-95 Sep-96 Mar-98 Sep-99 Mar-01 Sep-02 Mar Mar-95 Sep-96 Mar-98 Sep-99 Mar-01 Sep-02 Mar Nominal Effetive Exhane Rate (12-Month Perent Chane) 25 M3X (12-month perent hane) Mar-95 Sep-96 Mar-98 Sep-99 Mar-01 Sep-02 Mar Mar-95 Sep-96 Mar-98 Sep-99 Mar-01 Sep-02 Mar-04 Soure: Central Bureau of Statistis Central Bank of Kenya and staff estimates

22 20 where inf is CPI inflation; maize is the annual perent hane of the ross prodution of maize; 6 enery is the annual perent hane of the world enery ommodity prie index; neer is the annual perent hane of Kenya s nominal effetive exhane rate; money is the annual perent hane of broad money (M3X) The model is estimated usin ordinary least squares. First a eneral model enompassin all potential fators that may have affeted inflation and their las was estimated. Seond variables and las that were not statistially sinifiant were eliminated sequentially The results for the final speifiation are presented in Table II.3. The results suest that the rowth rate of broad money the world ommodity prie index of fuel and enery the nominal effetive exhane rate and the ross prodution of maize are important determinants of inflation. Speifially the results suest that: A one perentae point inrease in the rowth of broad money has been assoiated with an inrease in inflation by 0.34 perentae points with a half-year la; A one perentae point derease in the rowth of ross prodution of maize has been assoiated with an inrease in inflation by 0.03 perentae points with a half-year la; A one perentae point inrease in the world ommodity prie index of fuel and enery has been assoiated with an inrease in inflation by 0.05 perentae point with a quarter la; and A one perentae point depreiation of Kenya s nominal effetive exhane rate has been assoiated with an inrease in inflation by 0.06 perentae with a one-year la. 6 Maize is inluded in the reression owin to its sinifiane in the Kenyan diets. This variable is preferred to a more areate measure suh as the total ariultural prodution whih inludes a sizable amount of ommodities meant for exports suh as tea offee and hortiulture that do not arry sinifiant weihts in the CPI basket. 7 For eah variable inluded in the reression an Aumented Dikey-Fuller test rejets the hypothesis of a unit root. 8 Fisal variables have also been used initially but were later dropped beause of their statistial insinifiane.

23 21 Table II.3. Kenya: OLS Estimates of a Redued-Form Inflation Equation onstant money enery maize NEER Overall Inflation 4.52* 0.34* 0.05* -0.03* -0.06* (7.01) (5.55) (2.46) (3.61) (2.47) Number of quarters laed Number of observations: 39 R-squared: 0.66 Prob(F-statisti): 0.00 Soure: IMF staff alulations Note: Sampe period is Q1:1995-Q3: An asterisk (*) indiates the variable is sinifiant at five perent sinifiane level. The fiures in the parentheses are absolute t-statistis based on standard errors alulated usin Newey-West heterosedastiity and autoorrelation onsistent ovarianes 9. Movements of explanatory variables oupled with oeffiient estimates presented in Fiure II.3 suests that broad money rowth is a key fator underpinnin inflation. The 18 perent rowth in broad money in 1996 appears to have been the main ause of the 15 perent inflation in early Furthermore disiplined manaement of monetary areates was partly responsible for the low inflation durin Likewise the exessively loose monetary poliy sine the seond half of 2003 has ontributed to the reent inflation. 10. Fators other than money have also played an important role in determinin inflation durin the past deade. In partiular the double-diit inflation experiened in 2000 ourred durin a period of relatively moderate rowth in money; low rop prodution a weakenin urreny and a hih rowth rate of enery pries appears to have been the main ontributin fators to hih inflation durin the period. As reards reent inflation a poor harvest hih enery pries and a weakenin Kenya shillin were also main ontributin fators. E. Poliy Impliations 11. Lookin ahead the overridin objetive of monetary poliy should be to maintain prie stability. While some of the fators ontributin to hih inflation suh as the hih enery pries and the poor harvest of food rops were beyond the ontrol of the CBK the CBK ould have mitiated inflation by adoptin less expansionary monetary poliy. Aainst this bakround with a view to urbin inflation the monetary proram under the Fund s Poverty Redution and Growth Faility envisaes a ut in broad money rowth to 7.5 perent durin 2004/05 from 13 perent in the previous fisal year. 9 9 Growth rate is in terms of a onstant proram exhane rate.

24 22 Fiure II.3. Kenya: Impat of the Reression Variables on Inflation Atual Overall Inflation Total Impat of the Reression Variables Mar-96 Sep-97 Mar-99 Sep-00 Mar-02 Sep-03 0 Mar-96 Sep-97 Mar-99 Sep-00 Mar-02 Sep M3X NEER Mar-96 Sep-97 Mar-99 Sep-00 Mar-02 Sep Mar-96 Sep-97 Mar-99 Sep-00 Mar-02 Sep Enery Pries 2 1 Maize Prodution Mar-96 Sep-97 Mar-99 Sep-00 Mar-02 Sep Mar-96-1 Sep-97 Mar-99 Sep-00 Mar-02 Sep Mi Soure: Staff estimates. Note: The fiures show the impat of the variables on inflation. The impat is measured by the produt of the oeffiient estimates of the fators and their manitude.

25 23 III. ESTIMATION OF THE EQUILIBRIUM REAL EXCHANGE RATE FOR KENYA 1 A. Introdution 1. One important indiator of a ountry s external ompetitiveness is the real exhane rate. 2 Given the stron evidene of a positive link between export performane and eonomi rowth notably in East Asia fosterin a ompetitive real exhane rate is interal to Kenya s development objetives. 2. Sine the early 1990s Kenya has made onsiderable proress in liberalizin its trade and exhane rate reime. In the early 1990s Kenya removed apital ontrols and moved from a fixed exhane rate reime to a manaed floatin system with the U.S. dollar as the prinipal intervention urreny. Durin the same period Kenya also embarked on trade liberalization whih involved a redution in the number of tariff bands from 15 in 1990 to 4 in 1999 and a lowerin of the top tariff rate from 100 perent to 25 perent. However its urrent trade reime whih is rated 6 on the IMF s 10-point trade restritiveness index (with 10 bein the most restritive) is the most restritive reime amon the three members of the East Afrian Community Developments in nominal and real exhane rates sine 1980 are presented in Fiure III.1. Generally speakin the nominal effetive exhane rate (NEER) has shown a depreiatin trend durin the past two deades while Kenya s domesti pries have outpaed those of its tradin partners and swamped the nominal depreiation of its urreny resultin in an appreiated real exhane rate. The fiure demonstrates that the real effetive exhane rate (REER) volatility has diminished sine the shift to a manaed float. 4. This hapter examines Kenya s CPI-based equilibrium REER. 4 It identifies a lon-run ointeratin relationship between the real exhane rate and a number of explanatory eonomi variables durin Usin the estimated ointeratin equation an equilibrium real exhane rate path is alulated and ompared to the atual 1 This hapter was prepared by Kevin C. Chen (AFR). 2 External ompetitiveness has many other aspets that are not diretly aptured by the real exhane rate. For instane unit labor osts labor quality physial infrastruture judiiary soundness politial stability and overnane affet a ountry s ompetitiveness. 3 For details on Kenya s urrent trade reime see Chapter V. 4 The onept and measurement of an equilibrium exhane rate are a ontentious issue in the eonomis literature. In addition there are always drawbaks to the various approahes that have been employed by different analysts. Therefore the results of the eonometri analysis presented in this hapter should be interpreted with this aveat in mind. 5 The sample is trunated at 1980 beause REER data were only ompiled beinnin in 1980.

26 24 Fiure III.1. Kenya: Exhane Rates and Relative CPI Nominal Exhane Rates (1990=100) Trade-Weihted Relative CPI (1990=100) NEER (Left Sale) US/Shillin Rate (Riht Sale) Mar-80 Mar-82 Mar-84 Mar-86 Mar-88 Mar-90 Mar-92 Mar-94 Mar-96 Mar-98 Mar-00 Mar-02 Mar-04 Mar-80 Mar-82 Mar-84 Mar-86 Mar-88 Mar-90 Mar-92 Mar-94 Mar-96 Mar-98 Mar-00 Mar-02 Mar REER (1990=100) 50 REER (12-month perent hane) Historial Averae REER Mar-81 Mar-83 Mar-85 Mar-87 Mar-89 Mar-91 Mar-93 Mar-95 Mar-97 Mar-99 Mar-01 Mar Mar-80 Mar-82 Mar-84 Mar-86 Mar-88 Mar-90 Mar-92 Mar-94 Mar-96 Mar-98 Mar-00 Mar-02 Mar Soure: Information Notie System. Note: An inrease in an exhane rate index means an appreiation. An inrease in the trade-weihted relative CPI means Kenya's CPI inreases faster than its tradin partners.

27 25 data. The eonometri results suest that the urrent REER level is above the equilibrium level implied by eonomi fundamentals. 5. The hapter is oranized as follows: setion B presents a theoretial and eonometri framework used in estimatin the equilibrium REER. The results are presented in setion C and the hapter onludes with poliy impliations in setion D. Theoretial Bakround B. Model 6. A sinle redued-form approah is used to estimate the equilibrium REER. 6 Speifially the equilibrium REER is assumed to be a funtion of several fundamentals whih inlude: 7 Relative produtivity of the tradable setor This lassi Balassa-Samuelson effet assumes that while pries of tradable oods are equalized aross ountries inreased produtivity rowth in a ountry s tradable setor relative to its tradin partners will bid up waes in the domesti eonomy. Assumin that produtivity rowth in the nontradable setor is slower than the tradable setor pries of nontradable oods will have to inrease to ompensate for the hiher waes thereby resultin in a rise in the overall CPI and hene a real appreiation of the loal urreny. Export pries of tea and offee Given the prominene of tea and offee in Kenya s exports an inrease in the prie of these ommodities will tend to improve Kenya s terms of trade and appreiate the real exhane rate. 8 6 This approah is one of the most standard approahes used to identify the equilibrium REER for a variety of ountries today. For a detailed survey on various estimation method for equilibrium REER see MaDonald (1995) Montiel (1999) and Rooff (1996). 7 These are the variables that are typially used to estimate the equilibrium exhane rate for developin ountries. Some papers have also used fisal and external indiators whih were also initially inorporated in the analysis but were later dropped owin to either statistial insinifiane or non-robustness. 8 Commodity pries instead of the terms of trade are used beause most empirial studies in this area have found that ommodity pries are stronly ointerated with the real exhane rate while findin little link between the real exhane rate and the terms of trade. See for example Chen and Rooff (2002) MDonald (2002).

28 26 Openness to trade A more restrited trade reime is likely to appreiate the real exhane rate as trade barriers suh as tariffs tend to raise pries in the tradable setor thereby inreasin overall pries and hene the real exhane rate. Net forein assets (NFAs) Hiher NFAs are likely to be assoiated with a more appreiated real exhane rate. As disussed in Lane and Milesi-Ferretti (2000) a deline in the NFA position implies a rise in the home ountry s net indebtedness to the rest of the world. Therefore over the medium term the home ountry needs a more depreiated real exhane rate to ahieve a larer trade surplus required to servie the hiher debts. Conversely a stron NFA position implies that the ountry an sustain a hiher trade defiit that is assoiated with an appreiated real exhane rate. In addition the NFA position an also be used as a proxy for net apital inflows whih tend to appreiate the real exhane rate. Data 7. The followin data were used to estimate the equilibrium REER: Ariultural produtivity relative to the rest of the world sine Kenya s main exports are ariultural produts hanes in ariultural produtivity vis-à-vis omparator ountries were used to examine the Balassa-Samuelson effet. For this purpose Colombia Sri Lanka Tanzania and Uanda were seleted as omparator ountries beause of the prominene of tea or offee in their exports. 9 International ommodity pries of tea and offee alulated as the averae of the ommodity indies of tea and offee. Openness to trade measured by the ratio of the sum of exports and imports to GDP; 10 and The net forein assets of the bankin system measured by the ratio of NFA of the bankin system to GDP. 8. The followin observations an be drawn from the data on the explanatory variables used to estimate the equilibrium exhane rates: (Fiure III.2) Kenya s produtivity in the ariultural setor has delined relative to omparator ountries suestin a more depreiated real exhane rate; 9 Produtivity refers to labor produtivity alulated as the ariultural output per worker in the ariultural setor. Data were obtained from the World Bank s World Development Indiators. 10 While the IMF s trade restritiveness index may be a better indiator for openness the data are only available after the mid-1990s.

29 27 Fiure III.2. Kenya: Eonomi Fundamentals Underpinnin the Equilibrium REER Relative Ariulture Produtivity (In Loarithm) 5.4 Commodity Prie Index of Tea and Coffee (In Loarithm) Trend 4.8 Trend Mar-80 Mar-82 Mar-84 Mar-86 Mar-88 Mar-90 Mar-92 Mar-94 Mar-96 Mar-98 Mar-00 Mar-02 Mar-04 Mar-80 Mar-82 Mar-84 Mar-86 Mar-88 Mar-90 Mar-92 Mar-94 Mar-96 Mar-98 Mar-00 Mar-02 Mar Openness (Perent of GDP in Loarithm) 11.0 NFA (In perent of GDP) Trend Trend Mar-80 Mar-83 Mar-86 Mar-89 Mar-92 Mar-95 Mar-98 Mar-01 Mar NFA Mar-80 Mar-83 Mar-86 Mar-89 Mar-92 Mar-95 Mar-98 Mar-01 Mar Soure: Kenyan Authorities World Development Indiators and Staff Estimates. Note: The trend is obtained from the Hodrik-Presott filter.

30 28 The averae export pries of tea and offee have trended downward suestin a more depreiated real exhane rate; The eonomy has beome less open suestin a more appreiated real exhane rate; and The NFA position has inreased suestin a more appreiated real exhane rate. Methodoloy 9. The Dynami Ordinary Least Squares (DOLS) estimator developed by Stok and Watson (1993) is used to identify the ointeratin (equilibrium) relationship between the REER and the explanatory variables. Speifially the DOLS estimates the ointeratin relation by an ordinary least squares reression aumented by the first differene of the explanatory variables toether with their las and leads The equilibrium real exhane rate path is alulated based on the estimated ointeratin relation. Given that the explanatory variables have exhibited a hih deree of volatility to derive a proxy for the equilibrium values for these explanatory variables followin MaDonald and Rii (2003) the Hodrik-Presott filter is used to smooth out the short-term noise in the explanatory variables. The equilibrium path is then derived by substitutin these smoothed variables in the reression equation. C. Results 11. The eonometri findins presented in Table III.1 are as follows: A one perent inrease in ariultural produtivity vis-à-vis omparator ountries is assoiated with a 0.8 perent appreiation of the REER; A one perent inrease in the averae export pries of offee and tea is assoiated with a 0.3 perent appreiation of the REER; A one perent inrease in openness is assoiated with a 0.5 perent depreiation of the REER; A one perentae point inrease in the ratio of NFA of the bankin system to GDP is assoiated with a 3 perent appreiation of the REER. 11 Formally suppose X t and Y t are two non-stationary and ointerated stohasti proesses then there exists a θ suh that Yt θx t is stationary. The DOLS of Stok and Watson (1993) estimates θ by runnin the followin reression usin the ordinary least squares: p 0 + θx t + δ j j= p Y = β ΔX u. t t j + t

31 29 Table III.1. Kenya: Equilibrium (Cointeratin) Relation between the REER and the Eonomi Fundamentals (1) (2) Coeffiient Coeffiient Variable (absolute t-statistis) (absolute t-statistis) Constant (1.32) (2.01) NFA (4.12) (5.20) LN(AGRICULTURAL PRODUCTIVITY) (2.45) (4.00) LN(TEA_AND_COFFEE) (2.02) (2.44) LN(OPENESS) (3.48) (3.48) R Number of Observations Prob(F-statisti) Soure: Staff estimates. Note: In reression equation (1) the Dynami Ordinary Least Squares of Stok and Watson (1993) is used to estimate the equilibrium (ointeratin) relation between Kenya's real effetive exhane rate (REER) and the eonomi fundamentals (explanatory variables) usin quarterly data durin Equation (2) is a modified version of equation (1) with insinifiant las and leads omitted. The dependent variable is the loarithm of REER. The explanatory variables inlude: NFA (the net forein assets of the bankin system in Kenya as a perentae of the GDP); AGRICULTURAL PRODUCTIVITY (the loarithm of the relative ariultural produtivity index of Kenya relative to Sri Lanka Colombia Tanzania and Uanda); TEA AND COFFEE (the loarithm of the averae of the international ommodity prie indies of tea and offee); and OPENNESS (the loarithm of the sum of exports and imports as a perentae of GDP). The numbers in the parenthesis underneath the explanatory variables are the orrespondin absolute t-statistis based on standard errors alulated usin Newey-West heterosedastiity and autoorrelation onsistent ovarianes. 12. Fiure III.3 shows the atual REER and the estimated equilibrium path. The eonometri results suest that Kenya s atual REER appears to be more appreiated than suested by eonomi fundamentals.

32 30 D. Poliy Impliations 13. Fosterin a ompetitive REER is key to Kenya s objetive of promotin stron rowth and poverty redution. Poliies should therefore be direted to: Enhanin labor produtivity Maroeonomi and strutural reforms inludin poliies aimed at inreasin labor market flexibility are key to improvin produtivity and ensurin that wae adjustments are uided by produtivity hanes and ost of livin and not by other riteria. Reently the authorities have established a wae-settin mehanism for publi setor employees and will issue uidelines for the private setor to help alin wae inreases to produtivity ains. Liberalizin trade Greater openness is essential to promotin ompetitive eonomi onditions and to reduin supply osts. Given that Kenya s trade reime is the most restritive amon EAC members further trade liberalization is warranted. Allowin a more flexible nominal exhane rate The authorities should allow the nominal exhane rate to adjust freely to fully reflet eonomi fundamentals. In this reard forein exhane intervention should be restrited to smoothin short-run flutuations. Maintainin prie stability Given that the rise in Kenya s domesti pries relative to its tradin partners has played an important role in the real appreiation of the REER durin the past deade lowerin Kenya s domesti ost struture by maintainin disinflation would help to enhane Kenya s external ompetitiveness. In this reard monetary poliy should be uided by the overridin objetive of prie stability.