Asia-Pacific Economic Statistics Week Seminar Component Bangkok, 2 4 May Monthly flash estimates of Economic Growth In Georgia

Transcription

1 Name of author Levan Gogoberishvili (Mr.) Asia-Pacific Economic Statistics Week Seminar Component Bangkok, 2 4 May 2016 Organization National Statistics Office of Georgia Contact address 30, Tsotne Dadiani Str., Tbilisi, Georgia Contact phone Title of paper Monthly flash estimates of Economic Growth In Georgia Abstract Gross Domestic Product (GDP) is the most common indicator of economic growth. International methodology of the system of national accounts (SNA 2008) recommends only annual and quarterly calculations of GDP. The main reason is that most of the indicators needed to calculate GDP are available annually or quarterly but are not available on a monthly basis. Nevertheless user wants to get a basic idea about the country's economic performance much earlier than quarterly GDP will be published. Because of this, some countries, including Georgia, have implemented the experimental calculations of Monthly flash estimates of economic growth. National Statistics Office of Georgia publishes monthly press release on flash estimates of economic growth 30 days after the end of the corresponding month. Given that the quarterly Gross Domestic Product is published 80 days after the end of the corresponding quarter, the release provides timely information about the economic situation and is much in demand from customers. The paper describes the sources and methods used for compilation of monthly flash estimates in Georgia. It shows how the administrative data from the Ministry of Finance are used to calculate monthly turnover of value added tax payer's. The paper discusses the reliability of results compared to the quarterly GDP and shows how big the difference between the monthly and quarterly calculations is.

2 I. Contents I. Contents II. III. Introduction....3 Monthly flash estimates of Economic Growth In Georgia 4 A. Data Availability B. Estimation Non-financial corporations Other sectors...9 IV. Conclusion 10 V. References..11

3 II. Introduction Macroeconomic policy-making in real time faces the problem of uncovering what is actually happening to the economy. "Time series of real GDP and related estimates of the output gap are widely regarded as important predictors of future inflation and thus are relevant to the problem of inflation targeting" [1]. Estimates of GDP are produced quarterly with the first estimates in Georgia available 80 days after the end of the corresponding quarter. In many countries including Georgia monetary policy is set more frequently than quarterly and in order to do so policy makers need to estimate what is happening within each quarter. A range of monthly series is currently available giving indications of short term movements in output. As the only available information, they are already exploited in various ways: economic experts examine monthly turnover of value added tax payer's, exports and imports of goods in order to assess the state of the economy; academic researchers exploiting high frequency econometric techniques make use of one or other of these series as the best available proxy for a broader measure of demand or output. If these monthly data are to be used to draw inferences about the state of the economy as a whole, then it is desirable that there should be some formal procedure for grossing them up to represent the whole of GDP. Such a procedure is likely to produce estimates of GDP which are worse than those which might be produced by direct measurement. On the other hand, it would be more satisfactory than simply making an informal inference from whatever happens to be the latest numbers available. The flash estimate of real GDP growth has been published monthly by the National Statistics Office of Georgia (Geostat) since January The indicator is based on output rather than expenditure data. Output indicators are generally regarded as being better guides than expenditure indicators to short-term movements in GDP. The main purpose of this paper is to describe how an indicator of monthly GDP is constructed in Georgia. Section A describes the available monthly data sources. Section B presents the methods used for estimation of monthly GDP at current prices.

4 III. Monthly flash estimates of economic growth in Georgia A. Data availability The components of the output measure of real GDP fall into three broad categories: 1. Series for which data are already available; 2. Series for which there are indicators; 3. Series for which no monthly information is available. Geostat works with a decomposition of the output measure of GDP into five institutional sectors of the economy. Table 1 shows these sectors together with the availability of indicator variables. Table 1 Availability of data for estimation of monthly GDP Institutional Sector Economic activity Availability of Data/Indicators % of GDP Financial corporations Non-financial corporations Financial intermediation 100% data available Industry, Construction and Private services Turnover of VAT payers Non-observed output No indicators Agriculture No indicators Households Food processing No indicators Imputed rent of owner occupied dwellings No indicators Government Public services Number of employees Available only with delay NPISH Other non-market output No indicators

5 Financial corporations include: Central and commercial banks, insurance companies etc [2]. Output of non-financial corporations consists of declared and non-observed parts. Nonobserved output is generated mainly due to tax evasion. According to SNA 2008, "people living in dwellings they own are considered to be selling housing services to themselves" [3]. The rents recorded in the national accounts therefore include both the actual rents paid by tenants and imputed rents in the case of owner-occupiers. "The amount of the imputed rent is measured by the rents paid for comparable housing in a similar part of the country." Public services cover public administration and defense, health and education. Some components of health and education are marketed and therefore included in the sector of non-financial corporations. Nonprofit institutions serving households (NPISH) are units formed by groups of households in order to supply services to themselves or to other households on a noncommercial basis. NPISH are mainly operating in the sectors of health, education and other social and personnel services. Table 1 also shows that the importance of the different sectors can change markedly over a relatively short time period. Nevertheless, much the most important sectors are non-financial corporations and government. The areas where the data are relatively weak are also those which do not contribute very greatly to the overall variability of GDP. However, it is necessary to interpolate these variables for which monthly data do not exist. At this stage data content of the monthly flash estimate is around 53 percent of the total required for the calculation of quarterly GDP. B. Estimation The components of output fall into three categories in the calculation of monthly GDP. For financial corporations, the monthly GDP is calculated from the monthly data of National Bank of Georgia (NBG). For non-financial corporations, the Ministry of Finance (MOF) provides monthly turnover of VAT payer's, which is indicator variable for the regression model suggested by Chow & Lin (1971) [4]. For the remaining sectors, for which there are no obvious indicator variables available, in order to estimate monthly GDP various autoregression models are used. 1. Non-financial corporations Output of non-financial corporations is the largest contributor to the overall GDP growth. The system of tax administration in Georgia makes monthly reporting of sales obligatory for all VAT payer enterprises. An enterprise is VAT payer if its annual turnover (revenue from sales) exceeds GEL (national currency). VAT declarations must be submitted no later than 15

6 days after the end of the month. MOF provides to Geostat the preliminary data on VAT turnover after 16 days from the end of the month. The second set of adjusted data is available after 28 days from the end of the months, just two days before the publishing flash estimates of GDP growth. There are about active VAT payer enterprises each month. Business register of Geostat contains main economic activities of these enterprises according to NACE rev.1.1 classification. Consequently, more than 97 percent of total VAT turnover is classified by 2 digit codes of NACE rev.1.1. For approximately 3 percent of total turnover, generated in the newly established enterprises, the main economic activity is unknown. Such enterprises are classified together as "unknown activities". In the first stage of calculations turnover generated in agriculture, financial intermediation and public administration are deducted from the total VAT turnover. Monthly GDP for these activities are estimated separately and by subtracting their turnover we avoid double counting. Then, we focus our attention on trade services (50, 51, and 52 of NACE rev.1.1). Output of trade services is trade margin. However, turnover of trade services also consists of value of purchased goods for resale. In order to obtain value of trade margin for the current month regression model (Chow & Lin 1971) is used, which is based on turnover and trade margin for the previous quarters. The same procedure is applied to the electricity transmission services. The next stage of calculation is transformation of VAT turnover values from current to constant prices. Geostat produces monthly indexes of producer price (PPI) and consumer price (CPI). PPI's are available for industry, construction and transport services. For the remaining activities, the relevant CPI's are used. Appropriate price indices are assigned to VAT turnover at each two-digit codes of NACE rev.1.1. Table 2 shows the process of calculations for non-financial corporations. VAT turnover for the same month of the previous year is used for weighting in order to obtain real growth rate of turnover for the whole sector of non-financial corporations. There is strong positive correlation between real growth rates of output and turnover. We therefore constructed linear regression model in SPSS to calculate real growth of output. Another regression model is used to move from the real growth of output to the real growth of GDP. There is also strong negative correlation between real growth of VAT turnover and real growth of non-observed output.

7 Regression model with reliable indicators of turnover and actually paid value added tax is used for calculation of real growth rate of non-observed output. Table 2 Calculation of real VAT turnover growth rate for February, 2016 Nace code VAT turnover Feb_15 VAT turnover Feb_16 Price index VAT turnover t - 1 Real growth of VAT turnover, % Contribution to Growth, % Unknown Total

8 Table 2 displays: two-digit codes of NACE rev.1.1 (column 1), VAT turnover figures at current prices in millions of national currency for February 2015 (column 2) and February 2016 (column 3). It also shows prices indexes for each two-digit code (column 4) and VAT turnover at prices of the previous year (column 5 = column 3 / column 4). Despite the small share of "unknown activities" in total VAT turnover, the contribution of this sector to the real growth is very important. These "unknown activities" are newly established enterprises, which are not recorded in business register. Their main activities become known later. Table 3 shows the negative ratio of correlation between VAT turnover and non-observed output. Table 3 Correlation between VAT turnover and non-observed output VAT_turnover Non_observed_ output Pearson Correlation ** VAT_turnover Sig. (1-tailed).000 N Pearson Correlation ** 1 Non_observed_output Sig. (1-tailed).000 N **. Correlation is significant at the 0.01 level (1-tailed). Table 4 Regression model in SPSS Coefficients a Model Unstandardized Coefficients Standardized Coefficients B Std. Error Beta t Sig. 1 (Constant) VAT_turnover a. Dependent Variable: Non_observed_output Table 4 displays currently used regression model for calculating real growth of non-observed economy.

9 2. Other sectors Table 1 indicates that there are some sectors: Agriculture, Public services, Imputed rent of owner occupied dwellings, Processing of food in households and Services provided by NPISH, for which there are no obvious indicator variable available. We use pure autoregression for NPISH services and household processing. However, there is some additional information for Agriculture and Public services, which can improve the monthly estimates of GDP. There are two different estimation methods of monthly GDP in Agriculture. Their use depends on the reporting month. At the beginning of the current year (January April), there are no data on expected harvest of crops and livestock production. Therefore, regression model is based only on data from the previous years. Agriculture survey results for the first quarter are available from May. These data contain livestock production for the first quarter and the size of sown area for annual crops. Considering this data, the regression model is changed and new model is used from May to December. The number of employees in public agencies is a key indicator to measure the real GDP for this sector. Geostat conducts the monthly survey of public agencies. The results of the survey, which include number of employed persons in public agencies, are available within a month delay from the calculation of monthly flash estimates of GDP. For example, during the calculation of monthly GDP for February, data on public sector employment for January are used. For calculation of the next month, data on employment for the previous month is included in the updated regression model. Table 5 shows the real growth of monthly GDP for the whole economy, which is result of the weighting of all sectors. Table 5 Monthly flash estimate of real growth of GDP for February, 2016 Institutional sector Weights Growth rates of real GDP, % Non-financial corporations Households Government NPISH Financial corporations Total

10 IV. Conclusion Table 6 displays growth rates of real GDP according to (i) monthly flash estimates; (ii) the first publication of quarterly GDP and (iii) the final publication of annually adjusted GDP. Table 6 Growth rates of real GDP based on monthly, quarterly and annual data sources Period of time Monthly flash estimates (30 days) Growth rate of real GDP, % The first publication of The final publication of quarterly GDP (80 days) annual GDP (11 month) 2012Q Q Q Q Q Q Q Q Q Q Q Q Table 6 also shows that there are small differences between the results of monthly flash estimates and quarterly calculations of GDP. The greatest difference was recorded in the third quarter of 2012 and amounted to 0.6 percent. Our estimate of 6.9% growth in 2012Q3 was superseded by a first official estimate of 7.5 percent growth and a second official estimate of 7.7 percent growth. There are more significant differences between the results of monthly flash estimates and annual calculations of GDP. Annual data sources are more developed than shortterm indicators. Overall, we can say that the monthly flash estimates provide information about the current performance of the country s economy. Users receive benefit from timely information, which actually does not differ significantly from the next official releases. Nevertheless, in the release of monthly flash estimates, we inform users that the differences between these estimates and

11 the final results might be significant. Moreover, we try to improve monthly flash estimates by developing new monthly data sources. This year for example, we plan to calculate monthly exports and imports of goods at constant prices. It will enable us to combine the results of monthly flash estimates from production and expenditure side. These comparisons will highly contribute towards the development of monthly flash estimates of economic growth. V. References [1] James Mitchell, Richard J. Smith, Martin R. Weale, Stephen Wright and Eduardo L. Salazar (2004). An Indicator of Monthly GDP and an Early Estimate of Quarterly GDP Growth. National Institute of Economic and Social Research, UK. [2] The System of National Accounts (SNA 2008). UN, IMF, World Bank, OECD. [3] Lequiller, F. & Blades D. (2006), Understanding NATIONAL ACCOUNTS. OECD. [4] Chow, G. C. & Lin, A. L. (1971), Best linear unbiased interpolation, distribution and extrapolation of time series by related series, The Review of Economics and Statistics