A PRELIMINARY COMPOSITE ECONOMIC PERFORMANCE INDEX FOR TRINIDAD AND TOBAGO

Transcription

1 A PRELIMINARY COMPOSITE ECONOMIC PERFORMANCE INDEX FOR TRINIDAD AND TOBAGO UWI, St. Augustine Campus, Trinidad Dr. Dave Seerattan and Ms.Julia Jhinkoo

2 Outline of Presentation Aim and Motivation of Study Purpose of a Composite Index Literature Review Methodology Composite Index of Economic Performance for Trinidad and Tobago Leading Composite Index and its uses Conclusion and Proposed Further Research

3 AIM and MOTIVATION To develop a Leading Composite Economic Performance Index (LCI) for Trinidad and Tobago which is based on readily available high frequency data and therefore can be produced on a timely and consistent basis at lower cost. The proposed LCI could act as an early warning system that signals changes in the macro-economy, as well as providing insights into how the economy works on a more frequent basis.

4 Composite Indices & Desirable Properties A Composite Index is based on a grouping of factors combined in a standardized way, providing a useful statistical measure of overall market or sector performance over time. A well designed composite indicator should be based on the best available evidence; be designed with transparent structures and assessed using appropriate multivariate and sensitivity analysis, Saisana and Cartwright (2007).

5 Literature Review Internationally there is a wide cross section of studies done on composite indices. The use of CI s is used in advanced economies USA, EU, China CI s are relatively new to the Caribbean region with Barbados being the only country to have one in current use.

6 Literature Review Jordan and Howard (2005) Cotire et al (2007) Cragiwell and Moore (2008)

7 Methodology Development of the LCI for Trinidad and Tobago was based on the checklist proposed by the OECD which proposes ten steps: 1. Theoretical Framework 2. Data Selection 3. Imputation of Missing Data 4. Multivariate Analysis 5. Normalization 6. Weighting and Aggregation 7. Uncertainty and Sensitivity Analysis 8. Back to the data 9. Links to their indicators 10. Visualization of the results

8 Methodology Data Selection A CI is suppose to be a mirror the business cycle of the country, GDP is the best indicator. It was converted to monthly series, weighted by exports.

9 Methodology Data Selection Possible Leading/Coincident Economic Indicators for Trinidad and Tobago Economic Indicator Frequency Source Exchange Rate per US dollar Monthly CBTT Liquefied Natural Gas Production (cu m) Monthly CBTT Crude Oil Production (000 s Barrels) Monthly CBTT Crude Oil Refinery Throughput (000 s Barrels) Monthly CBTT Production of Motor Gasoline (000 s Barrels) Monthly CBTT Production of Gas and Fuel Oil (000 s Barrels) Monthly CBTT Future Gas Price (US$ per Million BTU) Monthly EIA Spot Gas Price Henry Hub (US$/MMBTU) Monthly EIA Future Oil Price WTI (US$ per barrel) Monthly EIA Spot Oil Prices WTI (US$ per barrel) Monthly EIA Production of Cement (tonnes) Monthly CBTT Production of Direct Reduced Iron ( 000 s tonnes) Monthly CBTT Production of Billets (000 s tonnes) Monthly CBTT Methanol Production (000 s tones) Monthly CBTT Inflation (Consumer Price Index per cent change) Monthly CBTT Net Official Reserves (US$M) Monthly CBTT Government Debt Outstanding (US$M) Monthly CBTT Central Government Overall Fiscal Balance (TT$000 s) Monthly CBTT Financial System: Loans Outstanding Consumers (TT$M) Monthly CBTT Financial System : Loans Outstanding Incorporated Businesses (TT$M) Monthly CBTT Commercial Banks : Real Estate Mortgage Loans (TT$M) Monthly CBTT Commercial Bank Ordinary Savings Deposit Rate (%) Monthly CBTT Commercial Bank Basic Prime Lending Rate (%) Monthly CBTT Commercial Bank Real Estate Mortgage Loan Rate (%) Monthly CBTT Composite Stock price Index Monthly CBTT/ TTSE Balance of Trade (US$M) Monthly CBTT Money Supply (M2) (TT$M) Monthly CBTT Index of Retail Sales Quarterly CBTT Unemployment Rate (%) Quarterly CBTT Index of Industrial Production Brazil, India and USA Monthly IFS Economic events (local and international) Monthly Author compiled from World News

10 Methodology Economic Event In trying to improve on the reliability of the proposed composite index, a new variable was considered. Economic events tend to influence the business cycle of the economy. Dominguez and Panthaki (2005) For this study we classified economic events as oil and gas discoveries, increases in minimum wage, budget speeches, elections, decisions of major Caribbean conglomerates and noteworthy international events.

11 Methodology Each country is unique and their economies are driven by different factors. Review of studies helped identified the proposed indicators for this study. The aim is to develop a composite index that is: 1. Reliable able to act as a good indicator of economic activity. 2. Consistent able to be produced within a timely and regularly period. 3. Econometrically Sound to be able to serve as a basis for short-term predictions of economic activity.

12 Methodology- Data Analysis The pre-selected indicators are examined and evaluated for their cyclical performance to validate their classification as a leading indicator. Each proposed indicator was tested using TRAMO to remove any outliers and seasonal factors that may be in the data. McGuckin et al (2007) recommends that the most recent data set be used and that indicators where data is missing be estimated. The Hodrick-Prescott filter is then applied to get the filtered times series of the various indicators.

13 Methodology- Composite Index Construction There are several methods that are used to develop a composite index: (1) The NBER Business Cycle Dating Approach (2) The OECD Methods of Composite Index (3) GDP Rule of Thumb (4) Peaks and Through of the Commerce Department business cycle indicators (5) Stock and Watson s Business Cycle Indicators (6) Markov Switching Models (7) Principal Component Regression (8) Principal Covariate Index (9) DI-AR-Lag Model which is based on the diffusion index (10) Dynamic Factor Model

14 Methodology- Dynamic Factor Model This method was chosen because it is has been proven to be most suited for data sets with a number of variables for a short time series, as is the case for this study. Dynamic Factor Models were originally proposed by Geweke (1977) as a time series extension of factor models which were developed for cross-sectional data; the main idea of the model which was proven was that a few factors can explain a large fraction of the variance of many macroeconomic series.

15 Methodology BUSY The BUSY program is used on the filtered series, it identifies : the turning points lagging, coincident and leading indicators The BUSY program makes available a selection of statistical techniques designed for conducting business cycle analysis on a possibly large set of time series. Two types of statistical procedures are offered. The first is an NBER-type of analysis that is based on descriptive statistics such as cross-correlations, coherences and phases of the cross spectra and Bry and Boschan dating procedure (see Bry and Boschan, 1971). The second is based on dynamic factor models, following the work by Forni et al. (1999, 2000). Both are aimed at building composite indices that are leading, coincident or lagging with respect to a reference series.

16 Methodology Evaluation phase Turning Point Identification

17 Methodology Relevance In determining the relevant variables for the CI the ratio common component variance over series variance is examined. A value 0.3 and below indicates that there is not strong a commonalty and the indicator is idiosyncratic in nature. Five variables were removed, and the test re-done to result in improved ratios.

18 Methodology Classification Classification of variables based on the Dynamic Factor Analysis Leading Series Coincident Series Lagging Series Balance of Trade Cement Production Composite Stock Price Index Brazil- Industrial Index of Production Commercial Banks and NFIs- Consumer Deposits Classifications of the proposed economic variables are based on the correlation behavior of the common parts of each variable with respect to that common part of the reference series. Exchange Rate Commercial Banks and NFIs Business Deposits Inflation Gas Fuel Production Events Iron Rod Production Central Government Debt Future Price of Gas Motor Gas Production Industrial Index of Production- India, UK, USA Gas Price Future Price of Oil LNG Production Central Government Fiscal Balance Oil Price Commercial Banks and NFIs Business Loans Refinery Oil Production Reserves Commercial Banks and NFIs Consumer Loans Retail Sales Index Commercial Banks and NFIs Mortgage Loans Interest Rate Loans Methanol Production Money Supply Mortgage Interest Rate

19 Lagging Index

20 Coincident Index

21 Leading Index

22 Composite Index of Economic Performance for Trinidad and Tobago

23 Composite Index

24 Turning Points Analysis of Indices TURNING POINTS OF INDICATORS (DYNAMIC FACTOR ANALYSIS ) BAL_TRADE T: P: T: P: T: P: T: P: T: BRAZIL_IIP T: P: T: P: T: P: T: P: CEMENT_PROD T: P: T: P: T: P: COM_STK_PRICE P: T: P: T: P: T: P: DEPOSIT_BUSINESS P: T: P: T: P: T: P: T: P: T: P: DEPOSIT_CONSUMER T: P: T: P: T: P: T: P: T: P: ER_US P: T: P: T: P: T: P: EVENTS P: T: P: GASFUEL_PROD P: T: P: T: P: GAS_FPRICE T: P: T: P: T: P: T: P: T: P: GAS_PRICE T: P: T: P: T: P: T: P: T: GDP T: P: T: P: T: P: GOVT_DEBT P: T: P: T: P: T: P: T: P: GOVT_FISCAL_BAL T: P: T: P: T: INDIA_IIP T: P: T: P: T: P: INFLATION T: P: T: P: T: P: T: P: T: P: IRON_PROD P: T: P: T: P: T: P: T: P: T: LNG_PROD T: P: T: P: T: P: T: P: LOANS_BUSINESS T: P: T: P: T: P: T: P: LOANS_CONSUMER T: P: T: P: T: P: T: P: T: P: LOANS_MORTGAGE T: P: T: LOAN_INTRATE T: P: T: P: T: P: T: P: T: P: METH_PROD P: T: P: T: P: T: MONEY_SUPPLY P: T: P: MORTGAGE_INTRATE P: T: P: T: P: T: P: T: P: MOTORGAS_PROD P: T: P: T: P: T: P: OIL_FPRICE P: T: P: T: P: T: P: T: P: T: OIL_PRICE P: T: P: T: P: T: P: T: P: T: REFIN_OIL_PROD T: P: T: P: T: P: T: P: RESERVES P: T: P: T: P: T: P: RSI T: P: T: P: T: P: T: UK_IIP T: P: T: P: T: P: T: USA_IIP P: T: P: T: P: T: P: T: P:7-2012

25 Out-of Sample Forecast

26 Conclusions We have demonstrated that a high frequency CI based on readily available data can provide accurate up to date information on current aggregate economic performance. To date the composite index that we have for Trinidad and Tobago can satisfy two of the three proposed criterion, it is reliable and consistent.

27 Conclusions cont d The preliminary forecast of GDP and the composite index signal a downturn in the economy of Trinidad and Tobago for the first three months of However it is not econometrically significant so it is not yet useful to make accurate predictions for economic activity for Trinidad and Tobago. Craigwell (2010), even it is impossible to utilise these indices to calculate with precision the influences of the many economic variables, they contribute nevertheless to the elaboration of the estimates of the future growth of the real economy.

28 Further Work To get generate a composite index that is more stable to within stand the econometric analysis. Variables to be added are 1. Number of companies registered 2. Number of building permits granted. 3. Tourism measure Compute CIs for the rest of the Caribbean economies.

29 Thank You for your attention.