Introduction. Descriptive evidence of the relationship between inflation and the cycle

Transcription

1 VARIATION IN THE CYCLICAL SENSITIVITY OF SPANISH INFLATION: AN INITIAL APPROXIMATION The auhors of his aricle are Luis Julián Álvarez and Albero Urasun, of he Direcorae-General Economics, Saisics and Research. Inroducion In he advanced economies as a whole, inflaion has in recen years shown considerable downward sickiness. This is paricularly sriking in ligh of he inensiy of he recession. Agains his background, several recen papers have analysed he relaionship beween he dynamics of prices and aciviy, providing a se of poenial explanaions for he downward sickiness of inflaion during his laes crisis [see, for example, IMF (2013) or Maheson and Savrev (2013)]. I has been poined ou ha he limied impac of he decline in aciviy on inflaion migh be indicaive of moderae oupu gaps. The recenly observed noable increases in unemploymen raes would largely be srucural in naure, so heir influence on wage bargaining and price-seing would be less han if hey were merely cyclical. Also highlighed is he fac ha he low raes of inflaion observed in recen years and heir scan variabiliy reflec he price-sabiliy sraegies pursued by numerous cenral banks, as is he case of he Eurosysem. In keeping wih his hypohesis, he anchoring of inflaion expecaions around moderae levels would explain small flucuaions in price variaions around reference values. Oher explanaions of he downward sickiness of inflaion poin o globalisaion, aribuing a greaer response by prices o he degree of slack in he world economy, wih less influence exered by he naional cyclical siuaion. Also, a lesser response o he cycle by business margins or, indeed, heir counercyclical naure, would also conribue o explaining he sabiliy of inflaion in a recessionary conex such as he presen. Lasly, inflaion rigidiy migh increase if companies opimise heir prices less frequenly when average inflaion falls. As a phenomenon, he reducion in he cyclical sensiiviy of inflaion has no been observed in all counries. For insance, he behaviour of he Spanish economy in he recen period has differed from his paern. Despie he upward momenum in prices linked o he fiscal consolidaion process (see he righ-hand side panel of Char 1), inflaion has declined o a greaer exen han in oher advanced economies (lef-hand side panel of Char 1). Specifically, he differenial wih he euro area in erms of he overall consumer price index has dipped from approximaely 1 pp on average in he period o close o 0.2 pp in he recessionary phase ( ), a reducion which has been even sharper in erms of core inflaion. Indeed, he cyclical sensiiviy of Spanish inflaion appears o have increased in recen years, which would be consisen wih some reducion in nominal rigidiies. In he curren seing, marked by weak domesic demand and he need o achieve gains in compeiiveness, his response is paricularly beneficial since he ensuing moderaion in prices is providing for momenum in ne exernal demand and sofening he decline in domesic demand. The aim of his aricle, which summarises he iniial findings of a more exensive projec ha analyses price-seing policies, is o provide empirical evidence on he changes in he response by Spanish inflaion o flucuaions in aciviy since he onse of he recession. Secion 2 presens a descripion of he relaionship beween various measures of inflaion and aciviy. Secion 3 ses ou he resuls of several economeric esimaes. And in he final secion, conclusions are drawn. Descripive evidence of he relaionship beween inflaion and he cycle This secion offers an analysis of simple correlaions beween differen measures of inflaion and of he economy s cyclical posiion or degree of slack. To correcly inerpre hese relaionships, i should be borne in mind ha he heoreical sign of he relaion- BANCO DE ESPAÑA 11 ECONOMIC BULLETIN, JULY-AUGUST 2013 VARIATION IN THE CYCLICAL SENSITIVITY OF SPANISH INFLATION: AN INITIAL APPROXIMATION

2 INFLATION DIFFERENTIAL AND MEASURES OF CORE INFLATION CHART 1 SPAIN/EURO AREA INFLATION DIFFERENTIAL YEAR-ON YEAR RATE OF SEASONALLY ADJUSTED CORE INFLATION 1. pp HICP Core HICP excl. energy CORE INFLATION CORE INFLATION EXCL. REGULATED PRICES AND TAXES SOURCES: Eurosa, INE and Banco de España. ship beween he inflaion rae and he cyclical posiion is no predeermined. On one hand, posiive (negaive) demand shocks end o raise (lower) prices and quaniies, generaing a posiive relaionship beween boh variables. Conversely, supply-side shocks or ax changes bear in opposing direcions on prices and quaniies, such ha he relaionship beween he wo variables is negaive. Accordingly, wo measures of inflaion are considered hereafer. Firs, he CPI, which includes componens paricularly affeced by supply-side shocks, such as energy and fresh food. And, second, a core inflaion measure, he CPI excluding energy and unprocessed food prices, which is much less sensiive o supply-side flucuaions. In addiion, in he case of his laer measure, he impac of price changes associaed wih changes in indirec ax and in regulaed prices, such as medicines, is sripped ou; accordingly, his measure may be expeced o be procyclical. This adjusmen has been paricularly significan in recen years, given he increases in VAT raes in 2010 and 2012, and dearer regulaed prices for cerain goods and services, such as medicines or he charges linked o various public services. Specifically, he downward rajecory shown by he CPI measure since mid-2011 conrass noably wih he upward course of he core inflaion measure (righhand side panel of Char 1). The macroeconomic lieraure on price-seing sresses he effec of he cyclical posiion on prices. Empirical analysis has o ake ino accoun ha he degree of cyclical slack in any economy is no direcly observable, making an approximaion advisable hrough consideraion of a broad se of measures. Specifically, his secion analyses four approximaions o he cyclical componen linked o he labour marke (year-on-year change in he unemploymen rae, unemploymen rae, recession gap 1 and cyclical unemploymen 2 ), and a furher four linked o changes in aciviy (GDP growh, change in consumpion, capaciy uilisaion and oupu gap). The analysis considers he expansionary period, along wih he period in which he Spanish economy faced a double-dip recession. Table 1 presens he simple correlaion coefficiens beween he differen measures of inflaion and cycle 1 Sock and Wason (2010) define his measure as he difference beween he curren unemploymen rae and he minimum rae in he pas 12 quarers (including he curren rae). 2 This measure is obained from an unobservable componens model ha enables he cyclical and srucural componens of he unemploymen rae o be separaed. A more deailed descripion of his mehodology can be found in Wason (1986). BANCO DE ESPAÑA 12 ECONOMIC BULLETIN, JULY-AUGUST 2013 VARIATION IN THE CYCLICAL SENSITIVITY OF SPANISH INFLATION: AN INITIAL APPROXIMATION

3 INFLATION AND CYCLICAL POSITION. CORRELATION COEFFICIENTS TABLE 1 Period: Period: Core in aion excluding regulaed prices and axes CPI Year-on-year rae Quarer-on-quarer rae Year-on-year rae Quarer-on-quarer rae Annual change in unemploymen rae Unemploymen rae Recession gap Cyclical unemploymen GDP Privae consumpion Capaciy uilisaion Oupu gap Annual change in unemploymen rae Unemploymen rae Recession gap Cyclical unemploymen GDP Privae consumpion Capaciy uilisaion Oupu gap SOURCES: INE and Banco de España. RELATIONSHIP BETWEEN INFLATION AND THE BUSINESS CYCLE CHART 2 CORE INFLATION AND YEAR-ON-YEAR CHANGE IN THE UNEMPLOYMENT RATE (a) Core in aion Unemploymen rae CORE INFLATION AND GDP Core in aion GDP LINEAR REGRESSION LINEAR REGRESSION LINEAR REGRESSION LINEAR REGRESSION SOURCES: INE and Banco de España. a Core in aion excluding regulaed prices and axes. considered. Broadly, i can be seen ha, in he recession, he inflaion measures are more procyclical han hey were previously. Along hese same lines, Char 2 depics dispersion diagrams of he relaionship beween he core inflaion measure (verical axis) and he wo measures of he degree of slack (horizonal axis) ha are o be used in he subsequen economeric analysis, disinguishing beween he wo sub-periods. Differen relaionships beween he variables can clearly be seen here. Specifically, during he expansion period, he higher demand or beer labour marke siuaion was no accompanied BANCO DE ESPAÑA 13 ECONOMIC BULLETIN, JULY-AUGUST 2013 VARIATION IN THE CYCLICAL SENSITIVITY OF SPANISH INFLATION: AN INITIAL APPROXIMATION

4 by higher inflaion. Conversely, he sharp conracion in aciviy in recen years has brough abou a change in his relaionship and he increases in he unemploymen rae or he declines in aciviy have been accompanied by a moderaion in inflaion. Some esimaes of he cyclical sensiiviy of inflaion To analyse inflaion dynamics more precisely, his secion presens esimaes of sandard Phillips curves. 3 According o his ype of relaionship, curren inflaion ( ) depends on e expeced inflaion ( ), on he cyclical slack in he economy (h ) and on an error erm (e ). Curren inflaion is normally higher if expeced inflaion increases and lower if he degree of slack in he economy increases. The cyclical sensiiviy of inflaion is deermined by he coefficien. Specifically he esimaed relaionship is: h e e Expeced inflaion is a variable which canno be observed, so assumpions have o be made abou is behaviour. In he esimaes presened, he approach of Ball and Mazumder (2011) is adoped. These auhors consider inflaion expecaions o be a combinaion of a forwardlooking componen and a backward-looking componen. The former is idenified wih he cenral bank s inflaion arge and he laer wih average inflaion in he pas year. For a quarerly frequency, he formula used o proxy inflaion expecaions is as follows: e 1 (1 ) ( ) The above expressions give rise o he equaion used o make he esimae: 1 (1 ) ( ) h e To isolae he impac of supply shocks, use is made of he measure of core inflaion (seasonally adjused) which excludes from he overall index he componens of energy and unprocessed food, as well as he effec of ax changes and regulaed prices. The degree of cyclical slack in he economy is proxied by he year-on-year change in he unemploymen rae or by he quarer-on-quarer rae of change of GDP. Table 2 shows he esimaes of his model of he Phillips curve. For he period , he higher aciviy gave rise o higher inflaion, alhough he effec is no saisically significan. Also, a subsanial backward-looking componen is esimaed in he price dynamics (1 ), so he inflaion exhibis noable ineria. The esimaion of he models for he period confirms he increased cyclical sensiiviy of inflaion in he mos recen period. Indeed, he coefficien of he measure of cyclical slack increases in absolue value and becomes significan. 5 This developmen would be consisen wih a decrease in he degree of nominal rigidiy in a recession. 6 3 For he sake of clariy, no versions of he Phillips curve for open economies are shown. The resul obained, namely increased cyclical sensiiviy, is no alered by he inroducion of impor prices. By conras, Ball and Mazumder (2011) repor lower cyclical sensiiviy in he Unied Saes for he mos recen period. 5 To avoid problems of simulaneiy, GDP growh is lagged by one period. 6 Using microeconomic price informaion, Klenow and Malin (2010) find ha he frequency of price adjusmens in he Unied Saes increased in he recession. BANCO DE ESPAÑA 1 ECONOMIC BULLETIN, JULY-AUGUST 2013 VARIATION IN THE CYCLICAL SENSITIVITY OF SPANISH INFLATION: AN INITIAL APPROXIMATION

5 PHILLIPS CURVE ESTIMATION TABLE 2 Esimaed coef ciens (p-value in brackes) Model wih year-on-year change in he unemploymen rae (0.10) -8 (0.5) 0.52 (0) (0) Model wih quarer-on-quarer rae of change in GDP 0.31 (3) 6 (0.13) 0.8 (0) 0.35 (0) SOURCE: Banco de España. ACTUAL VERSUS FORECAST INFLATION CHART 3 MODEL WITH YEAR-ON-YEAR CHANGE IN UNEMPLOYMENT RATE MODEL WITH QUARTER-ON-QUARTER RATE OF CHANGE IN GDP FORECAST (a) FORECAST (b) ACTUAL FORECAST (a) FORECAST (b) ACTUAL SOURCES: INE and Banco de España. a Dynamic forecass esimaing he model o 2007 Q. b On his evidence, firms would adjus downward heir prices o check he decline in profi resuling from subsanial falls in demand. Furhermore, he esimaes show a decrease in he ineria of inflaion. Char 3 shows he dynamic predicions obained using he aforemenioned models, beginning in 2008 Q1, and he observed behaviour of inflaion. If he coefficiens esimaed for he firs period are used, inflaion during he recession is lower han ha prediced by he model, perhaps due o an increase in cyclical sensiiviy in he more recen years. By conras, similar exercises for he Unied Saes show observed inflaion o be higher han would be expeced from hisorical relaionships, so here was a flaening of he Phillips curve. Anoher vial quesion is wheher i would have been possible o predic he behaviour of inflaion in he period if he change in he relaionship beween inflaion and aciviy had been known. The char shows he dynamic predicions from 2008 Q1 assuming ha he esimaed coefficiens for he second period are known. I makes i plain ha he models used have a relaively high explanaory power when i comes o accouning for he lower inflaion seen in he Spanish economy in recen years. An alernaive way of analysing he changes in he cyclical sensiiviy of inflaion is o esimae a version of he model in which he coefficien of cyclical slack is no consan bu BANCO DE ESPAÑA 15 ECONOMIC BULLETIN, JULY-AUGUST 2013 VARIATION IN THE CYCLICAL SENSITIVITY OF SPANISH INFLATION: AN INITIAL APPROXIMATION

6 ELASTICITY OF INFLATION TO THE MEASURE OF THE DEGREE OF SLACK CHART MODEL WITH YEAR-ON-YEAR CHANGE IN THE UNEMPLOYMENT RATE MODEL WITH QUARTER-ON-QUARTER RATE OF CHANGE IN GDP ELASTICITY ELASTICITY SOURCE: Banco de España. raher varies wih ime, like ha shown in Char. 7 I clearly indicaes ha he cyclical posiion has had a higher impac on price changes in he mos recen period. 8 Conclusions The las few years have seen a noable decrease in he inflaion differenial beween Spain and he euro area counries. This aricle presens evidence ha Spanish inflaion is currenly more sensiive o he high degree of slack in he economy han in he pas, which conrass wih he weaker relaionship beween hese variables observed recenly in oher advanced economies. This resul coninues o hold if differen economeric specificaions and differen inflaion and aciviy indicaors are considered. In he curren seing, his greaer cyclical sensiiviy of prices is promping a sharp correcion of he compeiive disadvanage buil up during he las expansionary cycle. This adjusmen is favouring he curren dynamism of he expor secor and miigaing he fall in domesic demand. However, o ensure ha he process of depreciaion of he real exchange rae coninues in he fuure, furher reforms will be needed o raise he compeiiveness of he markes for cerain producs and o enhance efficiency in he facor markes. This would enable he Spanish producive sysem o increase he comparaive qualiy of is goods and services, 9 and o bring is selling prices more ino line wih he coss borne and wih he specific condiions of he business cycle. A way of building on he research repored in his aricle would be o deermine he relaive imporance of he various facors deermining he cyclical sensiiviy of inflaion. Specifically, i would be of ineres o analyse possible differing responses of nominal wages, produciviy or he margins se by firms. Also, i is imporan o look a wheher his iniial research has idenified a change in he response of prices o flucuaions in aciviy or is findings are a consequence of possible asymmeries beween upurns and downurns in he relaionship beween hese wo variables This dynamic esimaor of slack was obained using he Kalman filer. 8 For purposes of comparison, models of he Phillips curve were esimaed recursively using samples wih a fixed number of years (specifically, five) and he resuls confirm he increase in he cyclical sensiiviy of Spanish inflaion in recen years. 9 Marín and Tello (2013) analyse he role of non-price compeiiveness as a deerminan of he expor aciviy of European firms. BANCO DE ESPAÑA 16 ECONOMIC BULLETIN, JULY-AUGUST 2013 VARIATION IN THE CYCLICAL SENSITIVITY OF SPANISH INFLATION: AN INITIAL APPROXIMATION

7 REFERENCES Ball, L. and S. Mazumder (2011). Inflaion Dynamics and he Grea Recession, Brookings Papers on Economic Aciviy, Spring. Inernaional Moneary Fund (2013). The dog ha didn bark: Has inflaion been muzzled or was i jus sleeping?, World Economic Oulook. April. Klenow, P. J. and B. Malin (2010). Microeconomic Evidence on Price-Seing, in B. M. Friedman and M. Woodford (ed.), Handbook of Moneary Economics, 3(6): pp , Elsevier. Marín, C. and P. Tello (2013). The expor aciviy and non-price compeiiveness of European firms, Economic Bullein, May, Banco de España. Maheson, T. and E. Savrev (2013). The Grea Recession and he Inflaion Puzzle, Economic Leers, forhcoming. Sock, J. and M. Wason (2010). Modeling Inflaion afer he Crisis, NBER Working Paper Series, Wason, M. (1986). Univariae derending mehods wih sochasic rends, Journal of Moneary Economics, 18, pp BANCO DE ESPAÑA 17 ECONOMIC BULLETIN, JULY-AUGUST 2013 VARIATION IN THE CYCLICAL SENSITIVITY OF SPANISH INFLATION: AN INITIAL APPROXIMATION

8