Inflation Expectations in Nepal

Transcription

1 74 ECONOMIC REVIEW Inflaion Expecaions in Nepal T. P. Koirala, Ph.D. There is a significan posiive relaionship beween inflaion and inflaion expecaions in Nepal, where he laer variable has been generaed under Adapive Expecaion Hypohesis (AEH). Using 33 annual observaions of acual inflaion from 1973 o 2006, one percen increase in inflaion expecaions has 0.83 percen impac on conemporaneous inflaion. The forecasabiliy of inflaion expecaions on curren inflaion is higher han ha of he expeced inflaion proxied by one-period lagged inflaion. The forecasabiliy of he model has been examined on he basis of minimum Roo Mean Squared Error (RMSE). Therefore, i is desirable for he policymakers o consider inflaion expecaions while formulaing moneary policy o anchor inflaionary expecaions of he economic agens. I. INTRODUCTION Inflaion expecaions play an imporan role in deermining macroeconomic variables of an economy. Undersanding he pah of inflaion expecaions is imperaive o he policymakers and economic agens in formulaing policies and economic decisionmakings. Conrolling inflaion hrough demand managemen has become a major objecive of moneary policy in he recen global conex. An emergence of various possible ransmission mechanisms of moneary policy has opened an avenue o scruinize he effec of inflaion expecaions on conemporaneous inflaion in he framework of anchoring inflaion expecaions o achieve he desired objecives of moneary policy. There is a posiive relaionship beween inflaion expecaions and acual inflaion (Dornbusch, Fischer and Sarz, 2000). Expecaion is a saemen abou an unknown fuure even (Frisch, 1983). Given he consan supply of goods and services in he marke, an excess of expeced inflaion over acual inflaion increases he consumers demand for goods and services leading o a rise in conemporaneous inflaion characerizing demand-pull inflaion. Similarly, given he fixed demand for goods and services, if he supply of goods and services by he producers decreases as a resul of an excess of expeced inflaion over he acual inflaion and hence a rise in conemporaneous inflaion, he inflaion is aribued o shor supply. If economic agens expec ha inflaion will ake place in he fuure, inflaion is sure o grow because he buyers would Research Deparmen, Nepal Rasra Bank, koirala@nrb.org.np

2 Inflaion Expecaions in Nepal 75 like o buy more and sellers wan o supply less. Such phenomenon is referred o as he self-fulfilling prophecy (Friedman, 1968). The producion of goods and services depends on he anicipaed or expeced inflaion. If he acual inflaion is higher han he anicipaed inflaion, producers increase heir producion paricularly in he shor-run. They do so because high inflaion is an incenive o he producers. A decrease in expecaions of fuure inflaion lead o a decrease in he curren inflaion via an incorporaion of reduced inflaionary expecaions ino curren wage and price agreemens beween employers and employees (Sargen and Wallace, 1981, Flood and Garber, 1980). Large swings in inflaion expecaions are he facors leading o high volailiy of oupu. II. LITERATURE REVIEW In he lae 1960s and early 1970s, expecaions of economic oucomes o he privae economic agens dominaed he field of macroeconomic analysis. The long-run policy prescripion as posulaed by Phillips Curve broke down since he early 1970s when he OPEC increased he price of crude oil. Consequenly, he world economy suffered from sagflaion and issues of inflaion expecaions gahered momenum. Friedman (1968) and Phelps (1967, 1968) based heir naural-rae heories of unemploymen on he Expecaions Augmened Phillips Curve, where he relaionship beween acual inflaion and unemploymen depended on expeced inflaion. They inroduced he inflaion expecaions in he erswhile Phillips curve using he hypohesis ha he workers as well as he firms are ineresed in he real wage rae raher han he nominal one (an absence of money illusion). 1 The Keynesian heory, which he Moneariss equae wih a simple Phillips Curve 2 wihou adjusmen for expecaions, canno explain he problem of acceleraed inflaion arising from high inflaion expecaions. Irving Fisher (1980) argued he imporance of expeced rae of inflaion while discussing real rae of reurn and expeced rae of inflaion as wo deerminans of nominal rae of ineres, ha is, i = c + % P, where, c is real rae of reurn and % P is change in inflaion rae. Ex pos, he price change will no be equal o is expecaion unless financial markes in he counry uilize all he available informaion; a possible risk premium, due o uncerainy abou he fuure price change, is incorporaed in nominal ineres rae conemporaneously (Giddy, 2000). Policy credibiliy, an issue which is gaining populariy a presen, is one of he major deerminans of inflaion expecaions. A lack of credibiliy of a policy would inhibi a sufficien fall in inflaionary expecaions (Fellner, 1976, 1979). In order o reduce inflaion expecaions, Jose, Slack and Sriram (2002) emphasized credibiliy, accounabiliy and ransparency in policy formulaion. Summers (1993) found he inverse 1 The price equaion for Augmened Phillips curve is e P & = P & β(u u n ) + v, where he inflaion rae has a posiive and one-o-one relaionship on he expeced inflaion rae & adverse supply shock and a negaive relaionship on he deviaion of he unemploymen rae from is naural rae. 2 P& = β( u un ) + v, where inflaion ( P & ) depends on he rae of unemploymen (u ) which diverge from naural rae of unemploymen ( u ), β sands for degree of responsiveness beween he variables and v sands for oher shocks affecing inflaion. n

3 76 ECONOMIC REVIEW relaionship beween cenral bank independency and reducion in inflaion expecaions. Nordhaus (1975) showed ha poliical business cycle inhibied he reducion of inflaion expecaions. Alesina (1987) and Kydland and Presco (1977) analyzed he prevalence of dynamic consisency problem arising from lack of credibiliy of economic policies o anchor he expecaions formaion of privae economic agens. Moneary and fiscal coordinaion is he panacea for policy credibiliy (Blinder 1982, Loewy, 1983 and Sargen 1986). However, credibiliy is very difficul o obain because of he inflaionary ineria (Croushore, 1992). Moreover, high credibiliy of moneary policy is o decrease inflaion expecaions which oherwise will increase oday s inflaion and ineres raes, and hence a reducion of real money demand (Baxer, 1985). Therefore, here is a negaive correlaion beween credibiliy and inflaion and ineres raes, and a posiive correlaion beween credibiliy and real cash balances and reserves. High inflaion in he pas combined wih indexaion and weak credibiliy of moneary policy can make price more sensiive o shocks and creae unsable inflaion expecaions (Crocke, 2000). The counries adoping inflaion argeing regime as a moneary policy framework rely on such regime as one possible means of sabilizing inflaion expecaions. Moneary policy credibiliy under he inflaion argeing regime sabilizes inflaion expecaions. Large swings in inflaion expecaions could be a facor causing high volailiy in oupu. A discreionary policy framework wihou announcing any numerical policy arges migh creae difficulies in forming inflaion expecaions. If moneary policy lacks credibiliy, he cenral bank is likely o influence expecaions hrough acions han hrough announcemens, and hence a reardaion of economic growh (Higo, 2000). As explained above, inflaion expecaion is an imporan variable ha conrols curren inflaion. Two heories are developed o explain he formaion of expecaions: he firs is he Adapive Expecaions Hypohesis (AEH); and he second is he Raional Expecaions Hypohesis (REH). If an economic agen has hisorical informaion on how he economic sysem funcions, expecaions are formed on he basis of adapive expecaions model where fuure values of a variable are relaed o he hisory of is pas values. Therefore, he AEH is a backward looking hypohesis. The curren expeced rae of inflaion according o he AEH is he weighed average of pas raes of inflaion where weighs decline geomerically as one goes back o he pas periods. The reduced form equaion o derive inflaion expecaions is = λ 1 + ( 1 λ) 1 where, he expeced rae of inflaion a ime is a weighed average of he acual inflaion and he expeced inflaion a ime -1, where he adjusmen parameers λ and ( 1 λ) serve as weighs. According o Frisch (1983), he weigh λ is fixed on he basis of memory of he economic agens. There are wo ypes of economic agens: shor memory economic agens and long memory economic agens. The shor memory economic agens are hose who base recen pas informaion in forming expecaions. In his case λ is fixed closer o uniy. I implies ha economic agens have only he las period informaion bu no he periods before ha and hence he expeced rae of inflaion a ime becomes he las period s acual inflaion. The long memory economic agens use all possible informaion hroughou he periods besides he recen pas informaion. In such a case, λ is fixed closer o zero. In pracice, however, i is difficul o find shor memory economic agens who base heir expecaions uilizing only las period s informaion in forming

4 Inflaion Expecaions in Nepal 77 expecaions. Therefore, in order o incorporae all he available informaion hroughou he pas, an applicaion of disribued lag model is used o find he appropriae λ value serving as weigh. The weigh paern for he differen values of λ yields he learning behavior of economic agens in he formaion of expecaions. 3 Raional Expecaions Hypohesis (REH) presumes ha expecaions formaion of a variable is deermined on he basis of economic heory ha deermines he variable (Sargen and Wallace 1973). According o Muh (1961) changes in he srucure of he economic sysem affec fuure expecaions of variables. The advanage of raional expecaions in expecaions formaion over oher mehods is ha i eliminaes he sysemaic forecas errors. However, i does no mean ha i eliminaes all he errors. In REH, condiional expecaion forecasing maers 4. In condiional forecasing, economic agens make probabiliy assessmens based on all available informaion a he ime of forecas. Mahemaically, i is represened as = E [ / I ] + ε where forecas of 1 is equal o he condiional expeced value of uilizing all available informaion a he ime of forecas, i.e. I 1. The error erm ε in condiional expecaion model, which is equal o E[ / I 1], should have zero expeced value and uncorrelaed wih any informaion available o economic agens. In oher words, he mean and variance of ε in he equaion ε = E[ / I 1 ] are zero and consan respecively. The empirical analysis of he models incorporaing expecaion variables dominaed he lieraure in he 1950s. A number of sudies aemped o search for he behavior and derivaion of expeced variables and is applicaion in he model. The adapive expecaions model was used o derive inflaion expecaions by Cagan (1956) in an analysis of hyperinflaion for Hungary. Using monhly daa from 1921 o 1924, Cagan found ha he demand for real cash balances is inversely relaed o he expeced rae of inflaion. Similarly, Nerlove (1958), using monhly daa ranging from 1921 o 1924 for Germany, also found an inverse relaionship beween demand for real cash balances and inflaion expecaions. He used he parial adjusmen model in he analysis. In Nepal, Khaiwada (1994) empirically analyzed he relaionship beween inflaion and inflaion expecaion as an addiional variable in a mulivariae inflaion equaion over he sample period for Nepal. Taking one period lagged rae of inflaion as a proxy for 3 Weigh when λ equals Pas Period Weigh Smoohing Shor Memory λ =0.9 Long Memory ( λ =0.1) -1 λ λ( 1 λ) λ( 1 λ) λ( 1 λ) n n 1 λ (1 λ) Ec. ec. 4 In such a forecasing mehod, expeced curren and fuure value of dependen variable is based no only on he pas value of he independen variable bu also on he forecased value of independen variable iself by he auxiliary model. The informaion from auxiliary model is included in he main forecasing model and hen he forecas is performed.

5 78 ECONOMIC REVIEW inflaion expecaion, he found an insignifican relaionship beween inflaion and inflaion expecaion because of he prevalence of unsable and unpredicable rae of inflaion. In such a siuaion, people may accord imporance o oher facors for expecaion formaion han depend solely on he raes of pas inflaion. III. APPROACH, OBJECTIVE AND METHODOLOGY The basic approach of his paper is o find he validiy of inflaion expecaion variable derived under AEH raher han he prevailing pracice of uilizing one period lagged inflaion as a proxy of inflaion expecaions in Nepal. The laer concep relies on he insananeous adjusmen beween acual and expeced inflaion where he speed of adjusmen beween acual and desired (expeced) inflaion is assumed o be equal o uniy. I ignores geomerically declining weighs of pas observaions in order o derive inflaion expecaions which demands furher research for he validiy. Keeping his in view, he objecives of his paper is o derive he inflaion expecaions series of inflaion for Nepal and o assess he relaionship beween inflaion and inflaion expecaions. This paper uses annual daa of naional urban Consumer Price Index (CPI) from 1973 o 2006 for he analysis. The firs difference of he logarihm of CPI gives inflaion raes. In he course of deriving he expeced inflaion, he AEH is aken ino consideraion. Expeced inflaion is no an observable variable and, hus, here is a need o deermine is values hrough an appropriae procedure. According o AEH, observaions of expeced inflaion are relaed o observaions of he pas. The model of adapive expecaions o generae inflaion expecaions is represened as: 1 = λ( 1 1 ) and is simplified form for he esimaion purpose is as: λ + 1 λ). 5 This equaion saes ha he = 1 ( 1 expeced inflaion ( ) a ime is he funcion of curren inflaion 1 and one period lagged forecas of inflaion 1 where adjusmen parameers λ and ( 1 λ) serve as relaive weighs given o each erm. The firs forecas value of he expeced inflaion series is proxied by he acual inflaion. n The minimizaion of he sum of Residual Sandard Error 2, is he ( RSE ) = = 1 e /( n 1) crieria for he opimal selecion of λ where e is residual error obained by subracing inflaion expecaions from acual inflaion. A rial and error search process is followed by fixing he value of λ ranging from zero o uniy in order o find he opimum desired value. In oher words λ ha yields he mos accurae forecas is he one ha achieves he lowes RSE. Afer generaing he inflaion expecaion series, a linear relaionship is examined beween inflaion and inflaion expecaion using bi-variae regression model. Since he 5 = λ 1 ( 1 λ) 1 2 n 1 n = 1 + ( λ 1) 2 + λ( λ 1) λ( λ 1) n + ( λ 1) n ; 0< λ <1 can be derived by aking lags and coninuous subsiuion yields. I simply saes ha he forecas in period is equal o a weighed average of all pas acual values and one iniial forecas.

6 Inflaion Expecaions in Nepal 79 variables used in his paper are in percenage change, ha is, firs difference of log of CPI, coefficiens are inerpreed as elasiciy coefficiens. The forecasabiliy of a model is examined by Roo Mean Squared Error (RMSE) of in-sample period forecas as well as n ex-pos forecas. The formula for he calculaion of RMSE is as: e 2. / n The model is seleced ha minimizes he RMSE. IV. RESULTS OF THE ANALYSIS In esimaing inflaion expecaion series for Nepal for analyzing inflaion and inflaion expecaion, he sudy uilizes he AEH. As derived in Annex 1 along wih he exercise presened in Annex 2, he value of weigh λ ha minimize RSE is 0.40 using model as λ + 1 λ). The weigh, ha is, λ can be considered as memory of = 1 ( 1 economic agens. If λ is close o zero, hen he weighs decline slowly and he economic agens have a long memory. The reverse is rue when λ is close o uniy. The weigh 0.40 gives an inference Diagram 1: Lamda minimizing RSE ha economic agens have FIGURE 1: Lamda Minimizing RSE neiher oo long nor oo shor memory in erms of inflaion expecaions in Nepal. The relaionship beween seleced 8.00 values of λ and RSE is 6.00 shown in Figure 1. Since he 4.00 value of λ equal o 0.40, he 2.00 economic agen (or sociey a 0.00 large) adjus inflaion expecaions wih some ime lags. RSE value Lamda Value If he acual inflaion and RSE expeced inflaion are ploed in a ime scale diagram, a smooh series of expeced inflaion is generaed, which is considered as long-run rend of inflaion, as shown in Figure 2. The expecaion gap is presened in Figure 3 depicing a convergence paern of errors. = 1

7 80 ECONOMIC REVIEW FIGURE Diagram 2: 2: Acual and Expeced Rae Rae of of Inflaion Inflaion Diagram 3: Expecaion Gap FIGURE 3: Expecaion Gap Inflaion Raes Years Inflaion Rae Years Acual Expeced Expecaion Gap Afer he selecion of inflaion expecaions series on he basis of minimum RSE, he series is used for esimaion. This sudy hypohesizes ha here is a posiive relaionship beween acual inflaion and expeced inflaion, ha is, changes in inflaion expecaions leads o a changes in acual inflaion. Mahemaically, = α + β ; β >0, where β is elasiciy of inflaion wih respec o inflaion expecaions. The effec of inflaion expecaions and one-period lagged inflaion on curren inflaion is presened in Table 1. TABLE 1: The Effec and 1 on (1973 o 2006) Equaion no (9.81) (8.05) Noe: Figures in he parenhesis are values. significan a 1 percen level. significan a 5 percen level significan a 10 percen level. DW In-sample RMSE Ex-pos RMSE The coefficien of expeced rae of inflaion possesses a priori sign, ha is, here is posiive relaionship beween inflaion and inflaion expecaions. The coefficien 0.83 is inerpreed as follows: one percen increase in expeced inflaion will have an effec of 0.83 percen increase in curren inflaion. The coefficien is saisically significan a 1 percen level. The R 2 value is no presened here because i is negaive and insignifican as he applicaion of he variables used in he model are in percenage change form. DW saisic is saisically significan a 5 percen level. Similarly, as shown in equaion 2, one percen increase in one-period lag inflaion will have an effec of 0.83 percen increase in conemporaneous inflaion. If a comparison is made beween he effec of

8 Inflaion Expecaions in Nepal 81 expeced rae of inflaion derived under AEH and expecaion under one-period lagged inflaion, he former explains he variaion of curren inflaion in a beer way. The in-sample RMSE and ex-pos forecas are examined o assess he predicabiliy of he models in Table 1. Using he las hree observaions o assess he ex-pos forecas of he model, he inflaion expecaion derived under AEH has beer forecasabiliy han ha of inflaion expecaion derived under one-period lagged inflaion. The ex-pos RMSE in case of he former is less han ha of he laer implying ha he expeced inflaion derived under AEH has higher forecasing abiliy han he expeced inflaion derived under oneperiod lagged inflaion. V. CONCLUSION This paper found ha here is a significan posiive relaionship beween inflaion and inflaion expecaions in Nepal. Using 33 annual observaions of expeced inflaion from 1973 o 2006, i is found ha one percen increase in inflaion expecaions will have 0.83 percen increase in conemporaneous inflaion. Since he inflaion expecaions derived under he AEH is he beer explanaory variable of curren inflaion as compared o inflaion expecaions under one-period lagged inflaion, he weigh used o derive inflaion expecaions under AEH minimizing RSE is found o be This weigh gives an inference ha he economic agens have neiher oo long nor oo shor memory in inflaion expecaions. The forecasabiliy of inflaion expecaions on conemporaneous inflaion, as inflaion expecaion calculaed on he basis of AEH, is higher han ha of he inflaion expecaions proxied by one-period lagged inflaion, where forecasabiliy of he model has been examined on he basis of minimum RMSE. Therefore, i is desirable for he policymakers o consider he impac of inflaion expecaions while formulaing moneary policy o anchor inflaionary expecaions of he economic agens.

9 82 ECONOMIC REVIEW ANNEX 1 : Derivaion of Inflaion Expecaions Series on he Basis of Differen λ Values (Weigh) Years CPI Log of log of Long Memory Economic Agens Shor Memory Economic Agens CPI CPI λ = 0 λ =.1 λ =.2 λ =.3 λ =.4 λ =.5 λ =.6 λ =.7 λ =.8 λ =.9 λ =

10 ANNEX 2 : Derivaion of Residual Squared Error (RSE) on he Basis of Differen λ Values (Weigh) Inflaion Expecaions in Nepal 83 Years CPI λ = 0 λ =.1 λ =.2 λ =.3 λ =.4 λ =.5 λ =.6 λ =.7 λ =.8 λ =.9 λ = n 2 e = = 1 n 2 RSE = e /( n 1) = = 1

11 84 ECONOMIC REVIEW REFERENCES Alesina, A Macroeconomic Policy as a Repeaed Game in a Two-Pary Sysem. Quarerly Journal of Economics 102: Baxer, M The Role of Expecaions in Sabilizaion Policy. Journal of Moneary Economics15: Cagan, P The Moneary Dynamics of Hyperinflaions. In M. Friedman (ed.), Sudies in he Quaniy Theory of Money. Chicago: Universiy of Chicago Press.. Crocke, A Moneary Policy Objecives in Ligh of Asian Crisis in Mahadeva, L. and G. Serne, eds., Moneary Policy Framework in a Global Conex, London: Rouhedge. Croushore, D Wha Are he Coss of Disinflaion? Federal Reserve Bank of Philadelphia Business Review, May-June. Dornbusch, R., S. Fischer and R. Sarz Macroeconomics. New Delhi: Taa McGraw-Hill Publishing Company Limied. Fellner, W Towards a Reconsrucion of Macroeconomics Problems of Theory and Policy. American Enerprise Insiue The Credibiliy Effec and Raional Expecaions: Implicaion of he Gramlich Sudy." Brookings Papers on Economic Aciviies 1: Fisher, I The Theory of Ineres. Reprined ediion. New York: Macmillan. Flood, R.P. and P. M. Garber An Economic Theory of Moneary Reform. Journal of Poliical Economy 88: Friedman, M The Role of Moneary Policy. The American Economic Review 8:1-17. Frisch, H Theories of Inflaion. Cambridge: Cambridge Universiy Press.. Giddy, I Global Financial Markes. Boson: Houghon Mifflin Company.. Higo, H Inflaion Expecaions in Japanese Moneary Policy. in Mahadeva L. and G. Serne (eds.) Moneary Policy Framework in a Global Conex." London: Rouhedge. Jose, A.S., Slack, G. L. and S. Sriram Saisical Principles for Inflaion- Targeing Regimes and he Role of IMF Dae Iniiaives. in Carson, C.S., C. Enoch and C. Dziobek (eds.). Saisical Implicaions of Inflaion Targeing:Geing he Righ Numbers and Geing he Numbers Righ. Washingon, D.C: IMF. Kydland, F. and E. Presco "Rules Raher han Discreion: The Inconsisency of Opimal Plans. Journal of Poliical Economy 85: Muh, J.F Raional Expecaions and he Theory of Price Movemens. Economica 29: Nordhaus, W The Poliical Business Cycle. Review of Economic Sudies 42: 1-16.

12 Inflaion Expecaions in Nepal 85 Nerlove, M The Dynamics of Supply: Esimaion of Farmers Response o Price. Balimore, MD: The John Hopkins Press. Okun, A.M The Poliical Economy of Prosperiy. Washingon, D.C.: The Brookings Insiuion. Phelps, E Phillips Curves, Expecaions of Inflaion and Opimal Unemploymen over Time. Economica 34: Phillips, A.W The Relaion Beween Unemploymen and he Rae of Change of Money Wage Raes in he Unied Kingdom, Economica 100: Sargen, T. J. and N. Wallace Some Unpleasan Monearis Arihmeic. Federal Reserve Bank of Minneapolis Quarerly Review 117: 1-17.