Indonesia: Selected Issues
Author: International Monetary Fund
Publication Date:
23 Aug 2006
eISBN:
9781451818376
Language:
English
Keywords:
ISCR; CR; consumer confidence confidence index; confidence index; balance sheet; coincident indicator; sales index; indicator model; candidate indicator; simple average; Consumption; Consumer price indexes; Private consumption; Consumer credit
Download PDF (518 KB)

The corporate sector in Indonesia has been recovering in recent years from the financial crisis of 1997–98. This paper analyzes the performance of the Indonesian nonfinancial corporate sector in recent years and discusses remaining challenges and vulnerabilities. The decline in corporate leverage may have resulted to a large extent from supply-side constraints. Indonesia was the country most severely affected by the Asian financial crisis, with GDP declining by 13 percent in 1998. Despite modest bank intermediation, bank financing has regained prominence as a source of corporate financing in recent years.

Abstract

The corporate sector in Indonesia has been recovering in recent years from the financial crisis of 1997–98. This paper analyzes the performance of the Indonesian nonfinancial corporate sector in recent years and discusses remaining challenges and vulnerabilities. The decline in corporate leverage may have resulted to a large extent from supply-side constraints. Indonesia was the country most severely affected by the Asian financial crisis, with GDP declining by 13 percent in 1998. Despite modest bank intermediation, bank financing has regained prominence as a source of corporate financing in recent years.

V. Developing Economic Indicators for Indonesia1

A. Introduction

1. The purpose of this paper is to develop and test coincident and leading indicators of economic activity for Indonesia. Research at Bank Indonesia (BI) is ongoing to develop indicators for overall growth trends using the OECD approach. But for now, the authorities and staff have tended to depend on several variables available on a monthly basis—for example credit growth, motor vehicle sales, cement production—to assess the current state of the economy and to conjecture where the economy is headed. However, these variables have often given mixed signals, which underscore the importance of identifying more reliable indicators for tracking and predicting overall economic trends. This analysis is intended to complement BI’s work by introducing a model-based approach for selecting better performing indicators.

2. Compared with industrialized countries, where indicators are often used to predict turning points in the business cycle, the objective of this paper is more modest. The aim is to construct composite indicators that focus on near-term forecasting and are simple enough to be easily updated every month, and then used by policy makers. Given the difficulty of forecasting overall GDP (see below), the focus in the paper is on developing indicators of private domestic demand, which constitutes about 85 percent of overall GDP.

3. The rest of the paper is organized as follows. Section B provides a brief overview of the literature on economic indicators. Section C describes the approach adopted for Indonesia, including data descriptions and basic clean-up procedures. Section D discusses selected coincident indicators for private consumption and investment and their in-sample performance. Section E covers leading economic indicators. Section F discusses out-of-sample forecasts and Section G provides some conclusions.

B. A Brief Overview of the Literature

Pioneering research

4. Burns and Mitchell (1946) initially pioneered the research on economic indicators for the US economy. Later, Moore and Shiskin (1967) added a formal weighting scheme by scoring variables in terms of their economic significance, statistical adequacy, cyclical timing, and business cycle conformity. Based on this method, the Conference Board in New York City currently produces leading and coincident indices for the US, and the OECD produces leading indicators for its member countries, as well as indices for six major country groupings.

5. This approach has some shortcomings, notably: (1) the lack of a theoretical basis establishing a relationship between indicators and activity; and (2) the use of an ad hoc weighting mechanism in constructing composite indices from individual indicators.

6. In response to the second criticism, Stock and Watson (1989, 1991) introduced time-series econometric analysis for deriving leading and coincident indices for the US economy. Their initial work utilizes dynamic factor models using a Kalman filter to estimate and evaluate the relationship between indicators and target variables. Subsequently, they introduced factor models with principal components (Stock and Watson (2002)), as these are easier to estimate, and can incorporate a much larger set of underlying variables than the models with the Kalman filter. Furthermore, the development of Markov-switching models by Hamilton (1989) and their application to business cycle forecasting introduced flexibility to let a model pick up potentially different relationships between activity indicators and a target variable for different phases of a cycle, which is an important advantage for a business cycle forecasting model. Altogether, these studies have set benchmarks for subsequent model-based analysis of economic indicators that attempt to capture and evaluate the relationship between indicators and target variables. Marcellino (2005) provides a comprehensive survey of the literature.

C. Strategy for Indonesia

Target variables for forecasting

7. The aim of this paper is to develop leading and coincident indicators for private consumption and investment. While in most studies the standard target variables are quarterly real GDP or monthly industrial production (IP), estimates of these for Indonesia showed a very poor fit (as shown by their low R-squares) and also had poor in-sample performance. The poor fit may reflect the weak quality of the historical GDP series–the series has a large statistical discrepancy due to data problems, including from the lack of timely statistics capturing government activities.2 The IP series reflects developments in only a small share of the economy, as secondary non-oil industrial production accounts for only 35 percent of GDP. Therefore, this paper resorts to forecasting the two major GDP components which account for 85 percent of overall GDP - private consumption, which constitutes about 65 percent of GDP and gross fixed investment, which is over 20 percent of GDP. In addition, these components better reflect underlying trends in private sector domestic demand than overall GDP, which, in some sub-periods, showed contrasting trends to private demand owing to strong government spending.

Type of forecasting

8. The paper focuses on identifying coincident and leading (two quarters ahead) indicators that explain movements in target variables for the whole sample under consideration. This is somewhat different from the OECD methodology that focuses more on identifying a turning point, and then searches for variables that predict turning points well. Moreover, predicting turning points generally requires long time series data that include several business cycles, as well as the ability to date peaks and troughs accurately, both of which would be difficult for an economy like Indonesia. For these reasons, several studies on emerging markets undertaken at the Fund (Simone (2001), Leigh and Rossi (2002), and Mongardini and Saadi-Sedik (2003)) chose to identify indicators that can effectively provide near-term forecasts of economic activity, rather than forecasting turning points.

Selection of indicators and time horizon

9. A reasonable number of monthly indicators are available for Indonesia via CEIC, although they differ in terms of the period for which they are available. The choice of indicators, therefore, involves a trade-off between having a wider variety of candidate indicators for a shorter horizon and having a smaller set of candidate indicators over a longer horizon. In the end, the shorter data set was used, as many indicators most relevant for assessing activity, including the retail sales index, the consumer confidence index, consumer credit, and business activity index, are only available from 2000. In addition, our preliminary analyses over the longer horizon (since 1993) confirmed a structural break in the series in 1997/98 due to the Asian financial crisis.

10. Monthly indicators representing both domestic and external developments were also used in the paper. Table A.1 provides the full list of available variables that were used in the paper and their starting dates.3 In addition, given that Indonesia is a relatively open economy, several indicators from developed markets were analyzed in order to capture developments in the world economy, an approach similar to Mongardini and Saadi-Sedik (2003).4 As for most of the nominal variables, both of the original nominal series and a real series deflated by headline CPI were tested.5

Table A.1.

List of variables

article image
1/

An empty cell indicates data start before January 1990.

Vt=α+βDEid–ul–Fitr,t+Ut(1)

Basic data clean-up: seasonal adjustment for Islamic holidays and unit root

11. Data were deseasonalized. Following Mongardini and Saadi-Sedik (2003), a combination of Census X12 and dummy variables for Islamic holidays is used to make seasonal adjustment. First, X12 is applied to all the variables. Second, each X12-seasonally adjusted variable is tested to detect the potential impact of “Eid-ul-Fitr,” corresponding to the end of Ramadan by estimating Equation (1). The summary of the result is given in Table A.2.

Table A.2.

Significance of Islamic Holiday Seasonality Effect and Results From Unit Root Tests 1/

article image
1/

The significance tests represent the statistical significance at 10 percent (*), 5 percent (**), and 1 percent(*). The results for Muslim holiday seasonality are shown only for monthly data, as none of the quarterly series show significant Muslim holiday effects.

2/

Significance with dummy variables for the Asian crisis period.

12. Islamic holiday seasonality is found to have a limited impact on the majority of the variables that were tested. However, statistically significant Islamic holiday seasonality is shown by production-related variables, including IP and cement sales. Figure 1 shows the X12 adjusted series and the X12 and Islamic holiday adjusted series for IP. In what follows, Islamic holiday adjustment is applied only for the variables that show statistically significant estimate for β in Equation (1).

Figure 1.
Figure 1.

Islamic Holiday Seasonality Example: Industrial Production (Index 2000 = 100)

Citation: IMF Staff Country Reports 2006, 318; 10.5089/9781451818376.002.A005

Sources: CEIC database and Fund staff calculations.

13. As in most studies involving time-series data, the different series are tested for unit roots. Table A.2 summarizes the result of the unit root tests. Most of the variables, except interest rates, do not reject the null hypothesis of a unit root. Therefore, first differenced data were used (after taking the logarithms of seasonally-adjusted series).

Estimation

14. In this paper, simple reduced form estimation models were used, similar to that in Mongardini and Saadi-Sedik (2003), to construct coincident and leading indicators. There were two main reasons for this. First, the attempt to use factor models with principal components (Stock and Watson, 2002) or using error-correction models (Simone, 2001) both presented difficulties given the limited number of series available for Indonesia and the shorter time horizon over which they are available.6 Second, reduced form estimation has several advantages. In particular:

  • Operational value: the model is simple enough to be updated every month and can be used to evaluate underlying trends and near-term prospects.

  • Intuitive interpretation: it can be ensured that the relationship between the target and explanatory variables are intuitively plausible. In contrast, dynamic factor models and Markov-switching models are often considered “black boxes.”

  • Providing benchmark results : the model identifies economically plausible and statistically significant indicators and evaluates their statistical performance.

D. Developing Coincident Indicators

15. Each candidate for a coincident indicator is screened and chosen on the basis of its relationship with a target variable. Following Mongardini and Saadi-Sedik (2003), the following model is estimated for each candidate indicator:

Δyt=α+βΔxt+u(2)ut=εt+θεt–1

where Δyt = quarter-on-quarter growth rate, measured as the difference of the logarithm of the target variable,

Δxt = standardized quarter-on-quarter first difference of the logarithm of a candidate variable, and

ut = error term with a moving average component MA (1).

16. A standard error and covariance matrix is estimated using the Newey-West heteroskedastic-consistent procedure. An indicator is chosen as a coincident indicator when the estimation shows a statistically significant coefficient for the variable, reasonably higher R-square compared to benchmark estimation only with constant and MA-term, and an intuitively plausible causal relationship with the target variable. In addition, the stability of the results was checked by estimating the model excluding the last 1-3 quarters.

17. Once promising individual indicators are identified, a composite coincident index is constructed using some or all of the selected candidate indicators. In the literature, composite indicators are considered to be superior to individual indicators, because they can cover wider aspects of the economy and are less affected by noise in individual series. The paper assesses both composite indices with equal weights and estimated optimal weights, and compares their performance.7

Coincident indicators for private consumption

18. The above procedure identifies three coincident indicators for consumption: consumer credit (nominal series), currency in circulation (nominal series),8 and the consumer confidence index. The following table summarizes the in-sample performances of the individual indicators in explaining variations in the real growth of private consumption. The table also shows two composite indices. For some models, the MA term is dropped as its coefficient is not significant.

19. Consumer credit explains about 40 percent of the variation in the growth of private consumption, with a statistically significant coefficient, followed by currency in circulation and the consumer confidence index. While the finding that consumer credit and the consumer confidence index are good indicators of private consumption needs little explanation, the relevance of currency in circulation suggests that a significant part of the economy remains cash-based.

20. Forecasting performance improves visibly when indicators are combined into a composite indicator. Composite indicator models have much better significance of coefficients and R-square. This finding is in line with other studies, which show composite indicators are superior to individual indicators (Leigh and Rossi (2002), for instance). Also, composite indices are more robust and the results do not change even when a few quarters are excluded from the estimating equation, while models with single indicators tend to show large changes in estimation results when a few quarters are excluded. Furthermore, although composite indices with estimated optimal weights (model 5 and 6 in Table 1)9 have slightly better R-squares, equal-weight models perform comparably well and indeed benefit from having a smaller number of explanatory variables, which is critical for small sample estimation. In addition, with optimal-weights models, multicolineality problems can arise during the estimation process.10 Moreover, weights can be strongly influenced by in-sample fit and the models might not have good out-of-sample performance. Therefore, use of equal-weight composite indices is preferable for operational purposes.

Table 1.

Coincident Indicators for Consumption

article image
Note: All the variables are seasonally-adjusted, transformed into logarithm, and first differenced. All the candidate coincident indicators are individually standalized. Each model is estimated by OLS, with Newey-West adjusted standard error and covariance matrix estimation. t-values in parentheses. *** indicates significance at 1 percent level, ** indicates significance at 5 percent level, and * indicates significance at 10 percent level.

Simple average of consumer credit and currency in circulation.

Simple average of consumer credit, currency in circulation, and the consumer confidence index.

Coincident indicators for investment

21. The same procedure identifies five coincident indicators: capital good imports, cement consumption, nominal total bank credit, nominal VAT revenue, and real M2. Most of these variables have a plausible economic relationship to investment. VAT revenue is correlated with investment to the extent that investment is stimulated by signs of stronger consumption, as reflected in higher VAT revenues. The following table summarizes estimates for individual indicators, some equal-weighted composite indices based on all or some of the above indicators, and optimal-weight models.

Table 2.

Coincident Indicators for Investment

article image
Note: All the variables are seasonally-adjusted, transformed into logarithm, and first differenced. All the candidate coincident indicators are individually standalized. Each model is estimated by OLS, with Newey-West adjusted standard error and covariance matrix estimation. t-values in parentheses. *** indicates significance at 1 percent level, ** indicates significance at 5 percent level, and * indicates significance at 10 percent level.

Simple averages of capital good imports and cement consumption.

Simple averages of capital good imports, cement consumption, and bank credit.

Simple averages of capital good imports, cement consumption, bank credit, VAT revenue, and M2.

22. Capital good imports explain about 70 percent of the variation in investment growth. 11 It seems that business sector demand has a high import content, and, as a result, business sector expansion in Indonesia is reflected in higher demand for imported capital goods for investment. The explanatory power of capital goods import is about the same or higher than any of the estimated equal-weight composite indices. Perhaps reflecting the dominant relationship between investment and capital goods imports, the explanatory power of a composite index decreases when it includes other variables that have a weaker individual statistical relationship with the target variable. The only model that performs better than the simple capital good imports model is the optimal weights model that uses capital good imports and cement consumption.

E. Developing Leading Indicators

23. The same methodology as that for coincident indicators is used to find leading indicators of activity. A model similar to Equation (2) is estimated with a lead of two quarters for the target variables (private consumption and investment), standard choice for leads with OECD indicators. The results are summarized in the following tables.

24. As for consumption, the procedure highlights the importance of the retail sales index as a leading indicator. Although the statistical relationship between any of the candidate indicators and the target variable is generally weaker than that for coincident indicators, as expected, the retail sales index can still explain about 40 percent of the variations in future consumption growth, and the estimation result is robust to the exclusion of up to two quarters (Table 3). It is somewhat unexpected to see the retail sales index as a leading indicator rather than a coincident indicator. However, the estimation results seem to indicate that the sales index covers consumption items that are more sensitive to changes in households’ purchasing patterns, and hence has some forecasting power over future overall consumption.

Table 3.

Leading Indicators for Consumption

(+ two quarters).

article image
Note: All the variables are seasonally-adjusted, transformed into logarithm, and first differenced. All the candidate coincident indicators are individually standalized. Each model is estimated by OLS, with Newey-West adjusted standard error and covariance matrix estimation. t-values in parentheses. *** indicates significance at 1 percent level, ** indicates significance at 5 percent level, and * indicates significance at 10 percent level.

Simple average of the retail sales index, the consumer confidence index present condition, and real 1 month SBI rate (-).

25. Unlike the analysis for coincident indicators, the fact that there is only one indicator that has a strong statistical relationship with future consumption cautions against drawing strong conclusions only from this one variable. This is particularly the case when the R-square of the leading indicator of consumption is not as strong as that of capital goods import and investment. Therefore, tracking a combined composite index including the retail sales index and a few of other economically plausible variables, in addition to a single indicator model, might be able to provide a cross check of the predictions. Table 3 also shows estimates with the consumer confidence index and the one month real SBI rate, as well as a composite index combining the three variables. Improvements in consumer confidence can stimulate households’ purchasing plans a few months ahead and increase consumption. An easing of real monetary conditions can reduce financing costs for big ticket items and stimulate near-term consumption. While individually these variables do not perform as well as the retail sales index in forecasting consumption, the composite index has a larger coefficient than the retail sales index alone and the R-square improves marginally.

26. The procedure identified six leading indicators of investment: OECD composite leading indicators (CLI) for Japan and the US, IP, the retail sales index, motor vehicle sales, and Jakarta Composite Index (JCI) in real terms (Table 4). Most of the indicators have a clear economic relationship to investment, covering production (IP), demand for final goods (the retail sales index, motor vehicle sales12), financial sector development (equity market index), and impact of developed economies (leading indicators for Japan and the United States). In fact, it is interesting to find CLIs for developed countries acting as leading indicators for investment. At a glance, activity in Indonesia does not seem strongly correlated with the rest of the world: the ratio of foreign investment in total investment is about 20 percent, and Indonesia’s export share in GDP is about 35 percent, which is low compared to the regional average of 60 percent.13 However, the results suggest statistically strong ties of Indonesian investment cycles to the developed countries.

Table 4.

Leading Indicators for Investment, two quarters ahead.

article image
Note: All the variables are seasonally-adjusted, transformed into logarithm, and first differenced. All the candidate coincident indicators are individually standalized. Each model is estimated by OLS, with Newey-West adjusted standard error and covariance matrix estimation. t-values in parentheses. *** indicates significance at 1 percent level, ** indicates significance at 5 percent level, and * indicates significance at 10 percent level.

Simple average of composite leading indicator for Japan and the US, industrial production, and the retail sales index.

Simple average of composite leading indicator for Japan and the US, industrial production, the retail sales index, motor vehicle sales, and JCI index in real.

F. Forecasting Performance of Indicators

27. In this section, out-of-sample performance of the indicators developed in the previous section is evaluated. However, the limited sample period prevents rigorous testing of out-of-sample properties. Forecasting performance is therefore assessed based on performance for the first quarter of 2006, using data up to end 2005. The panel chart shows the results for consumption and investment of coincident/leading indicators.

28. All the indicators track in-sample developments of the target series well. Furthermore, coincident indicators for consumption predicted the further deceleration of this aggregate in the first quarter 2006. Moreover, coincident indicators for investment predicted the pickup of investment.

29. In addition, leading indicators also predicted the direction of quarter-on-quarter growth for 2006 Q1. Although the fall in consumption and the pickup in investment were much sharper than the predicted values, the models performed reasonably well given the uncertainties following the sharp rise in oil prices. Based on data up to the first quarter of 2006, the leading indicators of both consumption and investment forecast a recovery starting Q3. This is consistent with gradually declining interest rates envisaged by BI for the second half of the year assuming inflationary pressures remain subdued.

G. Conclusion

30. This paper identified coincident and leading indicators of private consumption and investment in Indonesia. The estimation approach is simple, easy to update, and provides an intuitive interpretation regarding the state of the economy. Although the scope for testing out-of-sample prediction by the model is limited, the forecasting results for the first quarter of 2006 based upon data up to the end of 2005 are reasonably good.

31. Overall, consumption indicators (both coincident and leading) seem to be somewhat weaker than investment indicators in predicting the respective aggregates.

This may reflect the limited number of time series, the short time horizon over which they are available, and weaker coverage of consumption data.

32. The compilation of additional data that better reflect trends in the economy is also needed to improve the forecasting ability of the model. Some research projects in this direction are already underway in Indonesia. For example, BI has adopted the OECD package and continues to search for composite indicators by expanding the set of candidate indicators, including internally accumulated data. The Ministry of Finance has also just started to compile detailed VAT revenue data, which seems to have produced promising results in explaining consumption.

Figure 2.
Figure 2.

Performance of Indicators

(In percent, q/q seasonally-adjusted growth rate)

Citation: IMF Staff Country Reports 2006, 318; 10.5089/9781451818376.002.A005

Sources: CEIC database and Fund staff calculations.1/ Including consumer credit, currency in circulation, and the consumer confidence index with estimated optimal weights.2/ Including consumer credit, currency in circulation, and the consumer confidence index with equal weights.3/ Including consumer credit and currency in circulation with equal weights.4/ Including capital good imports and cement consumption with estimated optimal weights.5/ Including capital goods import, cement consumption, and bank credit with equal weights.6/ Including the retail sales index, the consumer confidence present condition index, and real 1-month SBI rate.7/ Including composite leading indicators for Japan and the US, IP, the retail sales index.8/ Including composite leading indicators for Japan and the US, IP, the retail sales index, motor vehicle sales and JCI index.

References

  • Burns, A. and W. Mitchell, 1946, “Measuring Business Cycles,” NBER Working Paper No.2.

  • Hamilton, J., 1989, “A New Approach to the Economic Analysis of Nonstationary Time Series and the Business Cycle,” Econometrica 57.

  • Leigh, D. and M. Rossi, 2002, “Leading Indicators of Growth and Inflation in Turkey,” IMF Working Paper WP/02/231.

  • Marcellino, M., 2005, “Leading Indicators” manuscript, Bocconi University.

  • Mongardini, J. and T. Saadi-Sedik, 2003, “Estimating Indexes of Coincident and Leading Indicators: An Application to Jordan,” IMF Working Paper WP/03/170.

    • Search Google Scholar
    • Export Citation

  • Moore, G. and J. Shiskin, 1967, “Indicators of Business Expectations and Contractions,” NBER Occasional Paper No.103.

  • Simone, A., 2001, “In Search of Coincident and Leading Indicators of Economic Activity in Argentina,” IMF Working Paper WP/01/30.

  • Stock, J. and M. Watson, 1989, “New Indexes of Coincident and Leading Economic Indicators,” NBER Macroeconomic Annual Report, pp. 351-394.

    • Search Google Scholar
    • Export Citation

  • Stock, J. and M. Watson, 1991, “A Probability Model of the Coincident Economic Indicators,” Leading economic indicators: new approaches and forecasting records, edited by K. Lahiri and G. Moore, Cambridge University Press.

    • Search Google Scholar
    • Export Citation

  • Stock, J. and M. Watson, 2002, “Forecasting Using Principal Components from a Large Number of Predictors,” Journal of the American Statistical Association, December 2002, Vol. 97, No.460.

    • Search Google Scholar
    • Export Citation

  • Stock, J. and M. Watson, 2004, “An Empirical Comparison of Methods for Forecasting Using Many Predictors,” manuscript.

1

Prepared by Hiroko Oura (EP, APD).

2

While revenue and expenditure data are available on a monthly basis for the central government, statistics measuring local government activities have a lag of two years.

3

Real sector and trade data are chosen from CEIC. Data are picked if they are available (1) on monthly basis and (2) with a few months lag (usually 2–3 months).

4

They included the U.S. and Euro area producer price indices and composite leading indicators from the OECD.

5

As discussed later in the paper, nominal series often produce more stable estimates than real series, which are often affected by noise in their deflator. For instance, the artificial spike in headline CPI in 4Q 2005 owing to domestic fuel price hikes distorted the deflated real series. In addition, volume data for trade were not used because: (1) import and export deflator estimation in Indonesia is known to be weak; and (2) there are longer lags with respect to the release of data on trade volumes.

6

Principal component estimates using 54 candidate variables (Stock and Watson, 2002, used 130), showed that more than 9 principal components would have to be included with data available only for 20 quarters this would not give meaningful results.

7

One benefit of the model-based approach over earlier “ad hoc” studies is to let the econometric model decide which indicators should be combined into a composite index and with what weights. However, more recent studies have found that model based composite indices are very similar to the equal weighted ones (Marcellino (2005)), and that the estimation of economic conditions are rather robust to the choice of method. In addition, estimated optimal weights might be strongly influenced by in-sample developments, especially when the sample is small, and produce relatively poor out-of-sample forecasts. Therefore, alternative indices are analyzed, some with optimal weights and others with equal weights, and compare their performances with our data.

8

Nominal series are preferred over real series, because the estimation results are strongly influenced by the jump in the headline CPI in the fourth quarter of 2005, owing to a large domestic fuel price hike and subsequent weakening in consumption in the quarter.

9

The estimated coefficients in a multivariable regression provide weights on each included indicator in a straightforward manner. For instance, with model 6, an optimal weight composite index is constructed by adding standardized consumer credit growth and currency in circulation growth after each of them are multiplied by their respective estimated coefficients.

10

Indeed, serious multicolinearity problem appeared with Indonesian data in preliminary analysis in which efforts were made to estimate optimal weight models following Mongardini and Saadi-Sedik (2003), by starting from a general model including majority of the variables, and eliminating one by one based upon the t-statistics of coefficients. Using principal components for estimation instead of raw indicators is one way to avoid this multicolinerity issue; however, as discussed in section C, a relatively large number of estimated principal components are needed to capture a reasonable portion of overall variations with Indonesian data.

11

The importance of capital goods imports as an indicator of investment was confirmed during a series of meetings with the authorities and the private sector participants. The strong relationship between capital goods import and investment was detected by varieties of methodologies, including simple correlation analysis and the OECD methodology (BI), with different data clean-up procedures.

12

It may appear implausible for motor vehicle sales to be a leading indicator for investment when it is not strongly related to private consumption. However, a large portion of local motor vehicle sales are of commercial, not passenger, vehicles. This means that car sales may reflect business demand more than private consumption.

13

Comparators include South Korea, Malaysia, the Philippines, and Thailand. The numbers are based on 2000–2005 average.

Same Series

Other IMF Content

Other Publishers

Cover IMF Staff Country Reports
Indonesia: Selected Issues
Author: International Monetary Fund
Volume/Issue:
Volume 2006: Issue 318
Publisher:
International Monetary Fund
ISBN:
9781451818376
ISSN:
1934-7685
Pages:
70
DOI:
https://doi.org/10.5089/9781451818376.002
  • View in gallery

    Islamic Holiday Seasonality Example: Industrial Production (Index 2000 = 100)

  • View in gallery

    Performance of Indicators

    (In percent, q/q seasonally-adjusted growth rate)

Islamic Holiday Seasonality Example: Industrial Production (Index 2000 = 100)

Performance of Indicators

(In percent, q/q seasonally-adjusted growth rate)