Sie sind auf Seite 1von 21

MEASURING INFLATION EXPECTATIONS THE INTERNATIONAL EXPERIENCE

George Kershoff (Editor): BER Pieter Laubscher: BER Andrie Schoombee: Department of Economics, University of Stellenbosch

Stellenbosch December 1999

EXECUTIIVE SUMMARY EXECUT VE SUMMARY


The results of inflation expectation surveys are mainly used in two ways, namely they serve as inflation forecasts and central banks use them to evaluate the credibility of their inflationfighting policies. The results of direct quantitative inflation expectation surveys are a better measure of inflation expectations than the results of qualitative surveys, such as that produced by means of the net balance technique. It is better to quiz various groups in society (such as business managers, households and financial market participants), as the difference in their expectations reveal important information on the future development of inflation. A survey of the practices of central banks that consult inflation expectation surveys reveals the following: Most central banks make use of at least three surveys, which are mostly conducted amongst business people and consumers. Broad-based surveys are far more commonly consulted than dedicated inflation expectation surveys. In all the surveys, the expected inflation figures or the percentage change in inflation (in terms of CPI) are required and not merely whether inflation is going to increase or decrease. Although the method used to gather the responses (postal questionnaires or telephonic interviews) and the size of the sample might have an influence, the response rates of the inflation expectation surveys are generally low in the order of 25 to 35 per cent. Quarterly surveys are more common than monthly surveys. Institutions release the results as they become available, but the results are generally made known at the same time in the case of surveys contracted specifically by central banks.

ii

TABLE OF CONTENTS TABLE OF CONTENTS


Introduction ........................................................................................................... 1 What is inflation targeting?..................................................................................... 1 How does a policy of inflation targeting work? ........................................................ 2 Reliance on forecasts ................................................................................... 3 Use of inflation reports................................................................................. 3 Allowance for flexibility................................................................................. 4 Where do inflation expectation surveys fit in? .......................................................... 4 How are inflation expectations measured? .............................................................. 7 Qualitative vs. quantitative measurement of inflation expectations ................................ 7 Inflation expectations of various groups..................................................................... 8 Are the results of inflation expectation surveys of use for econometric forecasting? ......... 9 How are inflation expectation surveys conducted internationally? ................................11 Conclusions regarding the measurement of inflation expectations ...............................15 Bibliography......................................................................................................... 17

iii

IINTRODUCTIION NTRODUCT ON
The South African Government has recently decided to adopt inflation forecast targeting as a framework for monetary policy. An inflation expectation survey is required to implement such a policy. The Reserve Bank has approached the Bureau for Economic Research (BER) to conduct such a survey(s) on behalf of the bank. In order to provide the basis for an informed discussion of the details of such a survey, the BER has reviewed the literature on the subject and contacted central banks that implement inflation targets. The results of these reviews and contacts are presented in this report. The following topics are dealt with below: What is inflation targeting? How does a policy of inflation targeting work? Where do inflation expectation surveys fit in? and How are inflation expectations measured? The above serves as background to evaluate options available to South Africa to measure inflation expectations. These options and their respective advantages and disadvantages are presented in the last section of the report.

WHAT IIS IINFLATIION TARGETIING? WHAT S NFLAT ON TARGET NG?


In recent years, there has been a growing trend toward the adoption of inflation targets as the primary focus for the conduct of monetary policy. Countries that have adopted official inflation targets in the 1990s include New Zealand, Canada, the United Kingdom, Sweden, Finland, Australia and Spain (IMF, 1996: 108). In 1998, the monetary policy framework of 55 of the 91 countries that participated in the Bank of Englands survey could be characterised as one of inflation targeting (Bank of England, 1999: 273). Inflation targets are designed to help the central bank achieve long-run price stability in three principal ways: by providing a nominal anchor for monetary policy, by improving the transparency and accountability of monetary policy, and by enhancing the central banks inflation-fighting credibility. During the 1980s and 1990s, a number of countries abandoned their traditional anchors, such as monetary aggregates and exchange rates. One reason was that the relationship of monetary aggregates to economic activity broke down in many countries, leaving those central banks that targeted monetary aggregates relying more on discretion

and looking at a wide range of information for guidance. Without an anchor, policy actions can drift under the influence of short-run economic disturbance and, in the process, monetary policy can develop an inflationary bias (Kahn & Parish, 1998: 6). Under an inflation targeting regime the dynamic of the policy implementation process results in shifts from multiple and shifting targets often under political control to a single, announced target with operational responsibility delegated to an independent central bank (Sherwin, 1997: 266). Explicit inflation targets play two main roles to reduce and control inflation: first, by communicating to the public the objective that monetary policy seeks to accomplish, they serve as a co-ordination device in wage and price setting processes and in forming the publics inflation expectations; and second, they provide a transparent guide to the conduct of monetary policy, whose commitment and credibility can then be assessed on the basis of whether policy actions are taken to ensure that targets are achieved (IMF, 1996: 108). For the public at large and the financial markets, it is arguable whether the explicit inflation target has been the key element in altering and anchoring inflation expectations, or whether it is the accountability and transparency commitments of the central bank that has mattered more (Sherwin, 1997: 265). According to Rudebusch and Svensson (1998: 1-2) inflation targeting is characterised by (1) a publicly announced numerical inflation target (either in the form of a target range, a point target, or a point target with a tolerance interval), (2) a framework for policy decisions which involves comparing an inflation forecast to the announced target, thus, providing an inflation-forecast targeting regime for policy where the forecast serves as an intermediate target and (3) a higher-than-average degree of transparency and accountability.

HOW DOES A POLIICY OF IINFLATIION TARGETIING WORK? HOW DOES A POL CY OF NFLAT ON TARGET NG WORK?
After comparing the monetary frameworks of four inflation targeting countries New Zealand, Canada, Sweden and the United Kingdom Lafrance (1997: 254) concluded that in each country the objective has been to maintain inflation, as measured by the percentage change in consumer prices, at low levels within a fairly narrow range. Governments have supported these objectives, either explicitly or tacitly. Policy actions are forward-looking and based on inflation projections, which range, in published form, from quite detailed (New Zealand and the United Kingdom) to broadly indicative (Canada). While the move to explicit targets has represented an important change in the monetary framework, the conduct of monetary policy, in terms of policy instruments and the understanding of the transmission process of monetary policy actions, has not changed radically in inflation targeting countries. While the

goal of achieving a target rate of inflation provides a framework for and discipline on decision-making, the day-to-day conduct of monetary policy in the four countries remains discretionary in nature. Inflation targets represent goals and not explicit rules for policy actions. Many inflation targeting regimes share common features, namely reliance on forecasts, the use of inflation reports and allowance for flexibility.

Reliance on forecasts
Given lags in the effects of monetary policy on inflation, central banks seeking to achieve a target for inflation need to forecast inflation and adjust policy to projected deviations of inflation from target. Specifically, monetary policy actions generally affect output and employment with lags of six months or longer and affect inflation with lags of 18 months and more. As a result, policymakers take action based on forecasts of inflation one to two years into the future. For example, if under the current setting of monetary policy instruments, inflation is projected to rise above target one year from today, policymakers might need to take action now to tighten the current stance of monetary policy (Kahn & Parish, 1998: 8). Policymakers use a variety of methods to forecast inflation. They can look at private forecasts, use information from financial markets (i.e. gauging inflation from long-term bond rates), surveys of inflation expectations and make projections based on various econometric models of the economy. Whatever the approach, a necessary condition for the successful use of inflation targets is that the central bank has some capability of forecasting inflation based on the current stance of monetary policy (Kahn & Parish, 1998: 8).

Use of inflation reports


Most central banks that target inflation regularly issue an inflation report. The purpose of the report is to explain what the central banks target is, describe how inflation has behaved relative to its target and indicate where inflation may be headed in the future. Some inflation targeting central banks actually publish their inflation forecast, as well as the risk surrounding that forecast. The central bank may also use the inflation report to explain why a target may have been missed and what actions, if any, might be necessary to bring inflation back to its target (Kahn & Parish, 1998: 8).

Allowance for flexibility


Given the difficulty of forecasting inflation and the likelihood that many economic shocks will have only temporary effects on inflation, all inflation-targeting regimes allow the central bank to sometimes miss its target. When such misses occur, however, the central bank usually has to explain why (Kahn & Parish, 1998: 8).

WHERE DO IINFLATIION EXPECTATIION SURVEYS FIIT IIN? WHERE DO NFLAT ON EXPECTAT ON SURVEYS F T N?
Changes in demand (such as a narrowing of the output gap) and supply conditions (such as a sharp increase in commodity prices) usually are the initial causes of a rise in inflation. This higher level of inflation, then, is mostly sustained by amongst other, inflation expectations. Consumers, trade unions, producers etc. build these higher inflation expectations into wage demands, asset prices and selling price. How are these inflation expectations formed? From a theoretical view point, inflation expectations are partially based on past values of inflation and partially on a rational assessment of all factors influencing inflation (see box below).

The formation of inflation expectations: adaptive vs. rational expectations


Are peoples inflation expectations simply a projection of past inflation (i.e. they rely exclusively on past levels of inflation to form an opinion of future inflation) or a rational prediction based on all information available? The adaptive expectation hypothesis (AEH) postulates that expectations of future inflation are based solely on some distributed lag of past values of inflation. Business people often rely exclusively on past data of the time series they predict, i.e. as if the past values of the indicator contains all the required causal information in order to predict future values. People often ignore information that is logically necessary to forecast the variable. Adaptive expectations are therefore often encountered in practice (Wolter, 1993). Proponents of the second view are adherents of the rational expectation hypothesis (REH). They believe people are rational and will make use of all available information to make forecasts or when they form expectations. In this way they will tend to eliminate any systematic errors, which could lead to bias in their predictions. Expectations are therefore efficient (in as much they are using all available information) and unbiased. The main argument against the adaptive expectations hypothesis is the restriction that expectations are formed by exclusively relying on the historic values of the time series concerned as if no other existing

knowledge or any rational thoughts on the matter are taken into account when expectations are formed. Gramlich (1983), for instance, found that money supply explains part of the variation in inflation expectations as represented by inflation expectation surveys. There are more studies finding statistically significant variables explaining the variation in survey expectations (see Gramlich, 1983). In other words, expectation formation does incorporate other existing information rather than only information on past inflation. Roberts (1998) concludes: survey expectations are not purely adaptive. The main critique against the rational expectations hypothesis, in turn, is the underlying assumption that everybody has the cognitive or statistical abilities to form rational expectations and that full use is made of existing information at the time of the expectation formation. Furthermore, no evidence exists that peoples behaviour intuitively approximates the laws of statistics or economic rationality. Most of the literature rejects the REH (see Gramlich: 1983). Roberts (1998) concludes: survey expectations are not purely rational either . Batchelor (1986) has a problem with researchers basing their conclusions on the analysis of qualitative expectation surveys. The problem is that qualitative surveys could be positively misleading in econometric testing of the nature of inflation expectation formation and can therefore not be used to ascertain the nature of inflation expectation formation. Evaluating direct quantitative expectation surveys, however, he finds a higher degree of rationality in expectation formation. The debate about adaptive and rational expectations is important when it comes to the prediction of turning points and changes in an underlying variable. Models based on the adaptive expectations hypothesis will always tend to miss the turning point with one or more periods while rational expectations models assume that people would revise or adjust their expectations according to the error made in the previous period. The idea of a learning process in expectation formation, therefore opens the door for auto-regressive elements, so that a truly rational expectations model is also of an adaptive nature (see Friedman, 1979: 32). The major conclusion from the literature regarding theories on expectation formation is that neither adaptive nor rational expectation hypotheses fully explain the formation of expectations. The adaptive and rational expectation hypothesis should be seen as two extreme models and that the truth should lie somewhere in between (Wolter, 1993: 5). Models incorporating both elements, i.e. an auto-regressive component (extrapolative, adaptive or regressive expectations) and a rational component (expectations as unbiased and efficient forecasts) perform best in explaining variations in expectations.

A survey of the literature on inflation targeting monetary policy frameworks and the inflation reports of a number of central banks reveal that the results of inflation expectation surveys are mainly used in two ways. 1. These results serve as inflation forecasts. Central banks use several forecasting techniques, such as econometric models, long-term bond rates and the results of inflation expectation
5

surveys. The literature search did not reveal whether (if any) of these banks uses the results of inflation expectation surveys as inputs in their econometric models. 2. Central banks use the results of inflation expectation surveys to evaluate the credibility of their inflation-fighting policies. A gradual narrowing of the gap between inflation expectations and actual inflation could indicate that the central bank convinced the general public and financial markets of lower inflation. This gap will only narrow gradually, as inflation expectations by nature have an adaptive and rational component. Furthermore, if a rise in the inflation-fighting credibility of the central bank leads to a reduction in inflation expectations and consequently actual inflation, the central bank would have been able to reduce inflation without having to resort to restrictive monetary policies, which would have lowered output and employment. If consumers and business believe the central bank is committed to achieving price stability, they will accept lower nominal wage increases, incorporate lower inflation and inflation risk premiums into asset prices and be more willing to make long-term commitments based on economic fundamentals instead of inflation expectations. This credibility effect can help reduce the output loss that typically accompanies disinflationary monetary policies (Kahn & Parish, 1998: 7). The issue of the credibility of monetary policy is crucial. If actual inflation and expected inflation is the same, then the presumption is that society reaches a better economic outcome. Multi-period or even single-period labour and debt contracts are based on inflation expectations. When expected and actual inflation are equal, then participants in contracts carry out the transaction as anticipated when the contract was signed. If a central bank is able to achieve its target rate of inflation and this target rate of inflation is the expected rate of inflation, then there are more periods when actual inflation equals expected inflation and society gets a better economic outcome (Johnson, 1997: 361). The central bank could put its credibility at risk if it does not adequately foresee changes in inflation. Central banks that adopted inflation targets, therefore, invest heavily in improving their forecasting capacity and technology (Sherwin, 1997: 271). If a central bank is worried that its policy suffers from low credibility (which is usually the case directly after the introduction of inflation targets), this ought to affect the speed with which the repo rate is cut for example. Drastic cuts can lead to an increase in inflation expec-

tations, affecting the room for further cuts. In this case the central bank may have to keep its policy interest rate higher than the forecast would seem to require (Heikensten, 1997, 287).

HOW ARE IINFLATIION EXPECTATIIONS MEASURED? HOW ARE NFLAT ON EXPECTAT ONS MEASURED?
QUALITATIVE VS. QUANTITATIVE MEASUREMENT OF INFLATION EXPECTATIONS
As a starting point, we have to distinguish between qualitative and quantitative measurement of inflation expectations. The so-called net balance statistic is the most commonly applied qualitative survey method. Survey respondents simply have to indicate whether a particular variable was up, the same or down compared to some reference period. By subtracting the proportion of the respondents that answered down from the proportion that answered up, a net balance value is obtained, which in theory should reflect the respondents underlying mean expectations. The Carlson-Parkin method of quantification is slightly more complicated. See Batchelor (1986) and Foster and Gregory (1977). In the case of the quantitative measurement of inflation, respondents have to supply the percentage change in inflation, mostly as measured by the CPI. Batchelor (1986) analysed the University of Michigans Institute of Social Researchs (ISR) quantitative and qualitative consumer survey results between January 1978 and January 1984. The quantitative mean of inflation expectations obtained from the direct quantitative survey of inflation expectations appears to be a more rational predictor of actual inflation compared to the results of the qualitative surveys. Qualitative inflation expectation surveys do track year-to-year movements in actual inflation, but not month-on-month changes. The net balance statistic is more erratic and correlates poorly with both the level and changes in quantitative measures of inflation expectations. Systematic errors occur at times, i.e. errors in one month are carried over to the following, when net balance statistics are used to estimate the mean inflation expectation of the population. Net balance statistics are also biased when inflation expectations or uncertainty is high. They tend to underestimate the level of inflation expectations and uncertainty, and vica versa (Batchelor, 1986). According to Batchelor (1986), better quantification techniques of qualitative survey results first need to be developed before they can be used as indicators of inflation expectations in an econometric model. Alternatively, it might be more useful to conduct direct quantitative inflation expectation surveys.

INFLATION EXPECTATIONS OF VARIOUS GROUPS


An important issue in the direct measurement of inflation expectations pertains to the responses of various response groups, e.g. households, businesses and professional forecasters. Much work has been done on the expectation formation of different groups in society and the results have been interesting and in some cases surprising. Gramlich (1986), for instance, found USA households average forecasts of inflation superior to those of professional economists. He analysed the results of the University of Michigan ISR consumer survey and the Livingstone survey of economists forecasts. Roberts (1998: 14) arrived at a similar conclusion. Furthermore, amongst consumers, the lower educated and poorer consumers inflation forecasts are superior to those of the higher educated and higher income consumers.

Source: Survey of Consumers, University of Michigan Englander and Stone (1989) show a further interesting result. An analysis of the inflation expectations of households (again the ISR consumer survey) and professional forecasters (results from the Decision Makers Poll and the Blue Chip Consensus in the USA) indicates that household inflation expectations differ in their impact from that of financial market participants and professional forecasters. Household inflation expectations have a strong bearing on

movements in nominal wages, while the inflation expectations of economists and financial market participants have a strong influence on interest rate determination. The fact that consumers will tend to build their inflation expectations into wage bargaining with employers, combined with the result that consumer expectation surveys explain much of future nominal wage growth, suggests that the survey is a reliable indicator of the underlying inflation expectations. Similarly, Englander and Stone (1989) find inflation expectations influence interest rate determination and that the financial market expectation surveys accurately reveal these expectations.

ARE THE RESULTS OF INFLATION EXPECTATION SURVEYS OF USE FOR ECONOMETRIC


FORECASTING? Two questions are pertinent to the usefulness or rationality of expectation surveys: 1. Do surveys contain forward-looking information, i.e. do they reveal new information and not just an extrapolation of the past? 2. Do expectation surveys reveal true expectations; i.e. do they succeed in measuring underlying expectations? To proponents of the adaptive expectation hypothesis, expectation surveys are of limited use since beyond information on past levels of the variable concerned they do not contain additional information, which could improve the rationality of the forecast. Given its major critique, i.e. the inefficient use of available information to predict, expectations surveys to assist in inflation forecasting should therefore be handled with utmost care (Wolter, 1993). Adherents of the rational expectation hypothesis also distrust expectation surveys, however, for a different reason. They argue that survey participants do not necessarily reveal their true expectations. Experience with business surveys, however, shows that the overwhelming majority of participants have a direct interest in the actual survey results. Respondents therefore have no interest in giving wrong answers (Wolter, 1993: 10). Englander and Stone (1989) also find that in three inflation expectation surveys conducted in the USA1, the surveys are accurate indicators of the respondents expectations upon which they tend to act.

The University of Michigan consumer survey; the Decision-Makers Poll amongst financial market participants and the Blue Chip Consensus, mainly amongst professional forecasters.
9

Analysts disagree widely on the usefulness of expectation surveys. Researchers even come to opposite conclusions using the same survey results. See Gramlich (1983) and Roberts (1998)2. It is clear that the various econometric tests applied to evaluate the rationality of survey expectations are very sensitive to the researchers assumptions. See an overview by Croushore (1997) of various tests on the Livingstone survey of professional forecasters. Most analyses of inflation expectation data reject the rational expectations hypothesis. A large component of inflation expectation formation appears to be backward looking; i.e. inflation expectations are to some extent shaped by past levels and changes in inflation. The fact that a large part of expectation formation in practice is dominated by an extrapolative or adaptive component, raises serious problems for the use of direct survey measures on expectation formation. However, Wolter (1993) concludes: surveys will continue to constitute an important means of obtaining a realistic and enhanced picture of expectation formation (Wolter, 1993: 18). Roberts (1998) concludes: I find that the surveys [of inflation expectations] reflect an intermediate degree of rationality: Expectations are never perfectly rational or as unsophisticated as simple auto-regressive models would suggest. Models incorporating both a distributed lag on past inflation and expectations data therefore fare best in predicting inflation. Expectation surveys do contain important future information. Its role in nominal wage growth and interest rate formation is, for instance, highly significant (Englander & Stone, 1989). Volatile and unpredictable movements in food and energy prices explain much of the bias and systematic error in inflation expectations. Inflation expectation surveys are better predictors of core inflation (Englander & Stone, 1989). This implies that when the primary sources of changes in inflation are related to normal business cycle and labour market forces, surveys perform better in predicting inflation. Croushore (1997: 8) makes the point that most of the problems with inflation expectation surveys developed during the 1970s and early 1980s when the oil price shocks led to significant bias and inefficiency in survey expectations of inflation. More recently, the Livingstone survey forecasts show less systematic error. Croushore (1993: 10) made similar conclusions regarding the Survey of Professional Forecasters (SPF).

The University of Michigan ISR consumer survey, inter alia, tracking households inflation expectations, are found to be an unbiased predictor of inflation over a 12 month period. However, the unbiasness deteriorates over the mid- to late 1980s (Englander & Stone, 1989). The Livingstone survey of professional economists (Roberts, 1998 and Gramlich, 1983) and the business survey run by the BEA (Gramlich, 1983) are found to be a biased predictor of inflation.
10

HOW ARE INFLATION EXPECTATION SURVEYS CONDUCTED INTERNATIONALLY?


The results of the survey of the international experience of countries where inflation expectation surveys take place are summarised in the table below. Information on the following is presented in the table: Which societal groups are targeted business people, professional forecasters, financial market participants, households or labour unions? Which institutions conduct the surveys private non-profit business interest groups, publicly financed research organisations or the central bank itself? What methodology is used are qualitative or quantitative questions asked per postal questionnaires, telephonic interviews or any other method? How are the questions phrased? What kind of surveys are used broad-based surveys of economic conditions and business attitudes that include one or more questions on inflation expectations or targeted, customised surveys that serve exclusively to determine inflation expectations? Furthermore, did the central bank specifically contract the survey to measure inflation expectations? If available, information is provided on the size and composition of the sample or panel of respondents, the response rate, the frequency of the surveys (monthly or quarterly), the cost and how the results are released. The methods used by five countries Canada, New Zealand, Sweden, Australia and the United Kingdom that adopted inflation targets and measure inflation expectations have been considered in depth. The information has been obtained from journal articles, inflation reports of the respective central banks and through contact with people at these banks. A number of countries Israel, Czech Republic, Chile, Spain and Finland that also implement inflation targeting have been left out due to a lack of information. Although the United States does not follow an inflation targeting policy, it has been included, as there are many references to their surveys in the literature. Furthermore, even though South Africa has not yet adopted inflation targets, the country has been included. The next section options to measure inflation expectations in South Africa refers to this information.

11

Country Canada

Target group Business organisations Professionals forecasters Households Professionals

Institution Conference Board (not for profit bus. org.) Conference Board Conference Board Statistics Canada AC Nielsen (private market research comp.) AC Nielsen

Methodology Questionnaire; expected % change in inflation in next 12 months Questionnaire; incl. no of economic indicators & % change in CPI over next 2 years Telephonic; incl. no of question on inflation expectations Questionnaires on 10 econ. indicators incl. inflation Questionnaires; current & expected monetary conditions, M3, CPI-X, BA, long bond, NZ$, trade weighted, GDP, current account, budget deficit, avg. hourly earnings and unemployment Telephonic; Q1: What is current rate of inflation; Q2: Is inflation expected to be higher, lower or the same in 12 months time; Q3: Expected actual inflation rate in 12 months time Questionnaires; inflation expectations over 12-month period; regional and sectoral breakdown; business survey type Bank employees visit companies in person to assess business mood Telephonic; expected % change in CPI over next one, 2 and 5 years; also expectations on wage, repo and exchange rate

Broad based / Targeted BB BB BB BB T

Sample size 550 Small no 1500 Small no 217

Response Rate 25% N/A N/A N/A Nov. 98 47% Aug. 99 37% 35%

Contracted by Centr. B No No No No Yes

Frequency Q Q ? M Q

Cost N/A N/A N/A N/A R129 000 pa N/A

Release of results First to on-line, then reports and then press

New Zealand

Private bus. enterprises Households

Respondents receive results before media and on web

1000 people in a random sample 1500 bank clients. Mainly financial market participants & business people 60 Panel of 350 Financial 90 Employer organ. 30 Employee organ. 30 Purchase managers in manuf. 100 Purchase managers in trade 100 Random sample of 2000 households 2000 manufacturers and service industries N/A

Yes

Business enterprises

Reserve Bank

60% but dropping

In-house

R320 000

Results of previous survey and new questionnaire are sent simultaneously to respondents

Business enterprises Business clients Sweden Different groups

Reserve Bank National Bank Statistics Sweden

T T

N/A 90%

In-house N/A Yes

Q Q Q

N/A N/A R298 000 Stats Sweden release results on predetermined and announced date; no prior access

Households Private business Financial market participants

Statistics Sweden National Institute of Economic Research Aragon Fondkommission (fund manager) and Prospera Research AB

Part of consumer survey, ask view on inflation one year ahead Business tendency surveys; has additional questions. Q1 Is inflation expected to be higher or lower? Q2 By how much is inflation expected to change? View on inflation 2 to 5 years ahead

BB BB T

N/A N/A N/A

No No No

M Q Q

Partly financed by EU N/A N/A

Published as annexure to labour force survey

12

Country Australia

Target group Households

Institution Inst. of Applied Economic & Social Research in Melbourne Australian Centre for Industrial Relations and Research

Methodology Telephonic; ask opinion on current economic climate. Q1 Expect prices to go up, down or remain the same in a years time. Q2 What % expect prices to rise/fall over this period? Telephonic; 10 questions are asked, Anticipated inflation rate in 6, 12, 18 and 24 months time; as well as wage expectations

Broad based / Targeted BB

Sample size Random sample of 1200 people Panel of union secretaries of top 22 trade unions

Response Rate 25%

Contracted by Centr. B Yes

Frequency M

Cost R223 000 p.a. R15 000 p.a.

Release of results

Labour Unions

BB

86%

Yes

Respondents receive summary of avg wage and inflation; detailed results not released

Professional forecasters United Kingdom Different groups

Reserve Bank Barclays Survey of Inflation Expectations (Basix)

Telephonic: inflation forecast for next 2 years Postal questionnaires; expectations of inflation 12 to 24 months ahead, RPI inflation except for general public, where measure is not specified

T T

15 priv. sector economists Gen. public, Trade Unions, Fin. directors, Bus. economists, Investment analysis Academic economists Manufacturers, Traders, Financial sector Service industry N/A

N/A N/A

In-house No

N/A N/A

N/A N/A -

Private business Private business Households Professional forecasters USA (does not have an inflation targeting monetary policy framework) Households

Confederation of British Industries (CBI) (Not for profit business organ.) BCC GFK and Mori Consensus Economics Survey of consumers at Inst. for Soc. Research, University of Michigan Conference Board, New York Federal Reserve, Bank of Philadelphia Livingstone Survey

Business tendency surveys; qualitative question on price changes; no questions on inflation expectations Business tendency surveys; no questions on inflation expectations Consumer confidence survey

BB

N/A

No

N/A

Internet & reports

BB BB

N/A N/A

No No

N/A N/A

Telephonic; expected % change in inflation over next one and 5 years; part of consumer confidence survey Telephonic; expected % change in inflation, part of consumer confidence survey Questionnaires; gathers quantitative forecasts of a range of macroeconomic variables

BB

500

N/A

Households Professional forecasters

BB BB

5000 50 - 90 economists large comp. (30%) Investment banks (30%), comm. banks (29%), academics (13%), labour, government & insurance (8%)

N/A 67%

Yes, but not part on inflation expectations No In-house

N/A

Internet, reports

M Jun & Dec

N/A N/A

Internet, reports -

13

Country

Target group Professional forecasters

Institution Federal Reserve, Bank of Philadelphia Survey of Professional forecasters (SPF) Blue Chip Economic Indicators (BCC) Drexel, Burnham Lambert Decision Makers Poll (DMP) Bureau for Economic Research (BER) MRA AC Nielsen for BER Andrew Levy & Ass. US Graduate School / Beeld Reuters

Methodology Questionnaires; levels, change and probability for GDP, inflation, etc.

Broad based / Targeted BB

Sample size 36 in1993

Response Rate N/A

Contracted by Centr. B In-house

Frequency Q

Cost N/A

Release of results -

Professional forecasters Professional forecasters South Africa (does not yet have an inflation targeting monetary policy framework) Private business Households Labour Professional forecasters Professional forecasters Fund managers

Questionnaires; range of macroeconomic variables

BB

Questionnaires; range of macroeconomic variables

BB

Sample size not known, portfolio managers, economists, managers 190 - 400 professionals 814 Manufacturers; 1050 Traders 2500 households 300 business enterprises 29 fund managers, private and public sector economists 53 domestic and foreign fund managers and private sector economists 22 Fund Managers

N/A

No

N/A

Anonymity of participants not protected -

N/A

No

N/A

Postal questionnaires; business tendency surveys; qualitative data; expected change in rate of increase of purchase prices Personal interviews; consumer survey, no questions on inflation expectations included Postal questionnaires; & change in avg. wage settlements Faxed questionnaires; % change in inflation and GDP one and two years ahead and one year ahead for other variables e.g. PCE, current account, exchange rate, gold price, R153 and deposit rate Faxed questionnaires; % change in headline CPI, core CPI, PPI, GDP, exchange rate, prime, BA and long term bond rate for next 6 quarters and 2 years Faxed questionnaires; % of portfolio invested in cash, bonds and equities; no questions on inflation expectations

BB BB BB BB

40% 46%

No No No

Q Q Q Q M

N/A N/A N/A N/A

Reports

N/A 100%

No No

BB

Sept 99 31% N/A

No

N/A

Fleming Martin

BB

No

N/A

14

CONCLUSIONS REGARDING THE MEASUREMENT OF INFLATION EXPECTATIONS


The results of direct quantitative inflation expectation surveys are a better measure of inflation expectations than the results of qualitative surveys, such as that produced by means of the net balance technique. It is better to quiz various separate groups in society (such as business managers, households and financial market participants), as the difference in their expectations reveal important information on the likely future development of a number of economic variables, such as inflation, wage increases and interest rates. The major critique of proponents of the rational expectation hypothesis is that inflation expectation surveys do not reveal respondents true expectations. However, research revealed that if certain conditions apply, inflation expectations do indeed reveal the true expectations of respondents. Some of these conditions include the following: A large and representative sample tends to reduce the possibility of sampling errors. Questions must be phrased neutrally to avoid bias in the survey responses. The timing of the survey is important. The questionnaire should be sent out so that all participants have the same information available, i.e. the survey answers should be returned before additional information on the variables surveyed becomes available. Anonymity could improve surveys ability to truly reflect respondents expectations, as it eliminates the publicity effect. The publicity effect refers to individual forecasters that tend to follow the consensus or tend to be bold in order to attract attention if the results are publicised. The major critique of proponents of the adaptive expectation hypothesis is that inflation expectation surveys do not reveal any new information. The same information can be gauged from a simple extrapolation of the past. However, research has revealed that the results of inflation expectation surveys do reveal extra information. The challenge is to extract this extra information and include it in an econometric model to forecast inflation. A survey of the practices of central banks that consult inflation expectation surveys reveals the following: Number of surveys and societal groups consulted: Most central banks make use of at least three surveys. An equal number of surveys are conducted amongst business people and consumers. These are also the most common groups to be surveyed. Surveys

15

undertaken of other groups, such as professional forecasters, labour unions, private sector economists and financial market participants, are less common. Institutions contracted and scope of surveys: Broad-based surveys are far more commonly consulted than dedicated inflation expectation surveys. Furthermore, all the central banks make use of at least one broad-based survey. Surveys conducted by outside organisations are far more common than in-house surveys. Only two of the five central banks, namely that of New Zealand and Sweden, that consult surveys done by outside organisations specifically contracted these organisations for this purpose. Methodology: In all the surveys, the expected inflation figures or the percentage change in inflation (in terms of CPI) are required and not merely whether inflation is going to increase or decrease. The business tendency surveys mostly make use of postal questionnaires. Most of the consumer surveys are conducted telephonically. However, in some countries, other societal groups, such as business people, are also surveyed telephonically. Telephone surveys are conducted over a period of 4 to 10 days. The numbers of respondents targeted in the telephone surveys, with the exception of the ACIRRT survey in Australia and the one done for the Swedish Central Bank by Statistics Sweden, are nearly double the numbers targeted in questionnaire surveys. Response rate: The response rates are generally low in the order of 25 to 35 per cent. No clear trend emerges when comparing telephone to questionnaire surveys, or when comparing surveys conducted with businesses to those conducted with consumers. It seems as though incentives, such as respondents receiving the survey results prior to release to the market, fail to bring about higher response rates. In New Zealand, the response rate is dropping over time, although respondents receive the results before being released to the market. Very high response rates (86% and above) are achieved with the more focused telephone surveys undertaken by Statistics Sweden and ACIRRT in Australia, even though no incentives are given for participating. Frequency: Quarterly surveys are more common than monthly surveys, in the vicinity of 2:1. The monthly surveys are exclusively consumer surveys done by telephone. Release of information: In the case of surveys contracted specifically by central banks, the survey results are generally made known to all interested parties at the same time. There are two exceptions: the survey conducted with businesses in New Zealand (respondents receive results first) and the one conducted by the Conference Board of Canada with businesses (on-line subscribers receive the results first, then subscribers to those reports and finally, the media).

16

BIIBLIIOGRAPHY B BL OGRAPHY
Bank of England, 1999. The use of explicit targets for monetary policy: practical experience of 91 economies in the 1990s. Quarterly Bulletin August 1999. London: Bank of England. pp. 272-281. Batchelor R A, 1986. Quantitative and qualitative measures of inflation expectations. Oxford Bulletin of Economics and Statistics 48 (2). May 1986. pp. 99-120. Croushore D, 1997. The Livingstone Survey: Still useful after all these years. Federal Reserve Bank of Philadelphia Business Review. March/April 1997. Englander A S and G Stone, 1989. Inflation expectations surveys as predictors of inflation and behavior in financial and labor markets. Federal Reserve Bank of New York Quarterly Review 14 (3). Autumn 1989. pp. 20-32. Foster J and M Gregory, 1977. Inflation expectations: the use of qualitative survey data. Applied Economics 9. pp. 319-329. Friedman B M, 1979. Optimal expectations and the extreme information assumptions of rational expectations models. Journal of Monetary Economis 5. pp. 23-41. Gramlich E M, 1983. Models of inflation expectations formation. A comparison of household and economist forecasts. The Journal of Money, Credit and Banking 15 (2). May 1983. pp. 155-173. Heikensten L and A Vredin, 1998. Inflation targeting and Swedish monetary policy experience and problems. Sveriges Riksbank Quarterly Review 4. Heikensten L, 1997. Inflation targeting the Swedish experience. Paper delivered at the Bank of Canadas 1997 conference Price stability, inflation targets and monetary policy www.bank-banque-canada.ca/english/res/cn97e-6.htm. pp. 281-295. International Monetary Fund (IMF), 1996. World Economic Outlook. October 1996. Washington DC: The Fund. Johnson D, 1997. The credibility of monetary policy: International evidence based on surveys of expected inflation. Paper delivered at the Bank of Canadas 1997 conference Price

17

stability, inflation targets and monetary policy www.bank-banquecanada.ca/english/res/cn97e-6.htm. pp. 361-370. Kahn G A and K Parish, 1998. Conducting monetary policy with inflation targets. Economic Review 83 (3). Federal Reserve Bank of Kansas City. Third quarter 1998. pp. 5-32. Lafrance, R, 1997. Background paper: The Monetary Frameworks of four inflation-targeting countries. Background paper to the Bank of Canadas 1997 conference Price stability, inflation targets and monetary policy www.bank-banquecanada.ca/english/res/cn97e-6.htm. pp. 245-260. Roberts J M, 1998. Inflation expectations and the transmission of monetary policy. October1998. Washington DC: Federal Reserve System. Rudebusch, G D and L E O Svensson, 1998. Policy Rules for Inflation Targeting. February 1998. Working Paper of the Sveriges Riksbank. www.riksbank.com. Sherwin, M, 1997. Inflation targeting The New Zealand Experience. Paper delivered at the Bank of Canadas 1997 conference Price stability, inflation targets and monetary policy www.bank-banque-canada.ca/english/res/cn97e-6.htm. pp. 261-273. Wolter S C, 1993. The use of survey results in respect to auto-regressive expectation formation. Paper presented at the 21st CIRET conference held by the Bureau for Economic Research at Somerset West.

18

Das könnte Ihnen auch gefallen