The development of the Indonesian service sector: a first analysis of the estimates

The housing sector in Indonesia’s official national accounts

Estimates for the period 1950-2000 are based on the value added estimate of 2000, which is extrapolated based on the number of households in Indonesia. This constant price series is inflated by the housing component of the consumer price index to arrive at current price estimates. In this way, we obtain estimates which are consistent with our estimates for the colonial period. Moreover it is an improvement of the housing estimates as made by BPS assuming housing to be a fixed percentage of GDP.

2.5 Financial sector

According to SNA 93 output of financial intermediation companies in banking, insurance services and pension fund services cannot be directly measured since such companies do not normally charge their customers for their services except for some minor incidental services. Banks earn their main source of income by the difference between the interest earned by providing loans and the interest paid on deposits. Pension funds and insurance companies accept contributions and invest them in order to pay their customers. Their output has to be measured indirectly. This has also been done in Indonesia.

The banking system in the Netherlands-Indies, 1900-1940

In the Netherlands-Indies a number of major banking institutions were active. These were the Java Bank, which functioned more or less as a central bank, a few large commercial banks (i.e. Nederlandsche Handel-Maatschappij, NHM; Nederlandsch-Indische Escompto Maatschappij, NIEM; and Nederlandsch-Indische Handelsbank, NIHB), the Government Pawnshop Service, the savings banks and the Government Credit System consisting of volksbanken (district or regency banks), desabanken (village banks) and desalumbungs (rice credit-banks).

This division is maintained in estimating the value added of the financial sector in the Netherlands-Indies between 1900 and 1940. The procedure adopted is the following. For the ‘big four’ the financial report of the Java Bank is taken as starting point. Whereas for the others annual reports are actually available, they have a number of short-comings: 1) They are often less detailed, 2) it is often hard to subdivide income in the Netherlands from income in the Netherlands-Indies, and/or 3) it is hard to subdivide income from banking and income from trade activities.

Therefore it is chosen to use annual reports by the Java Bank, which give a very detailed statement of income and expenditures. Value added of the Java Bank is then divided by the share of credit extension of the Java Bank in total credit extension of the ‘big four’.

For the other financial institutions, i.e. the Saving Banks, the Government Pawnshop service and the Government Credit banks, detailed financial accounts are available, which make it possible to estimate value added in a direct way.
Estimation methods for the financial sector, 1950-2000

In the 1950s data on profit and wages for banks and insurance companies was directly taken from the firms. From banks scattered over Indonesia input from 70 banks was obtained, while from insurance companies only a few provided information. But based on information from the tax agency estimates could be made about the income of these insurance companies (Neumark 1954, Muljatno 1960).

Input-output tables usually act as ‘anchor’ for future estimates, but also for revisions of earlier national accounts (Van der Eng, 2005, p. 245). For the period 1950-2000 we will therefore take the estimates in the I-O tables as starting point. The intermediary years will be extrapolated weighted by total credit extension and inflated by the consumer price index.

3. Assessing the reliability of the estimates of value added in the service sector
An important part of this paper relies on the reconstruction of the service sector within the framework of national accounts. Now that we have briefly discussed the methods and sources used, it might be helpful to shortly assess the reliability of these estimates.

Whereas it is rather difficult to make a precise estimate of the margin of error, it is possible to make a qualitative statement on the reliability for the individual series. Since the wider availability of detailed statistics on the different (service) sectors one would assume that the official estimates from BPS, at least since the 1970s, have a lower margin of error than those for the colonial period. Therefore the year 1930 is chosen as a benchmark to assess the reliability of the estimates between 1900 and 1960. This margin of error is most probably higher than the one for the BPS estimates and can therefore be considered a minimum reliability.

Following Horlings (1995, p. 103-104) and Smits (1995, p. 60) the classification of Feinstein (1972, p. 21-22) is used to divide service industries according to the quality of the estimates. Each service sector is assigned a specific margin of error. The reliability of the sector as a whole is calculated by weighting these margins based on the percentage share of each industry in total tertiary value added in 1930.

The results of this procedure for Indonesia can be found in table 5. Railways, air transport and communication are all classified in category A, because the necessary data could be taken directly from the source. Moreover, calculation of the product does not require too many manipulations.

Table 5: Quality of Indonesian service sector estimates, 1900-1960
	A	B	C	D
Transport
Rail	x
Road				x
Water		x
Air	x
Communication	x
Trade
Foreign		x
Domestic			x
Government		x
Housing			x
Financial sector		x
Others				x

Notes: quality of estimates (following Feinstein, 1972, p. 21-22; Horlings 1995, p. 103-104; Smits, 1995, p. 60):

A: Excellent, margin between 1% to 5% C: Fair, margin between 15% and 25%

B: Good, margin between 5% to 15% D: Weak, margin between 25% and 50%
In category B water transport, foreign trade, government and the financial services are grouped. For these industries information is relatively abundant, and the additional information provides a large enough coverage to state that the resulting picture is quite reliable. At the same time, however, some rather bold assumptions have to be made. Examples are KPM’s contribution to total value added in water transport, the development of the wage sum of the central government, and the ratio between value added and credit extension in the financial services.

Estimates for domestic trade and housing are considered to be fairly reliable. For these sectors there is sufficient data to make estimates for benchmark years, although requiring some additional estimates and assumptions. The development of trade margins for example is assumed to remain constant. Furthermore it is difficult to disaggregate the development of house rents to different types of dwellings.

The least reliable estimates are the ones for road transport and the sector ‘others’. The estimates for road transport are partly based on a survey on taxis in 1929 in Surabaya, because no other information is available. This estimate was extrapolated to other years based on the number of cars. Estimates of traditional road transport are based on the number of workers in this industry as found in the 1930 census and the average wage of an unskilled labourer. These are rather arbitrary procedures and therefore these estimates have to be taken with some caution.

If these margins are weighted on the basis of the percentage share of each industry in total tertiary value added in 1930 the result is a margin of error between 15 and 25 per cent. This seems a very acceptable margin of error. Horlings, for example, estimated the reliability of his service sector estimates for the Netherlands, 1800-1850 to be between 15 to 30 per cent, whereas Smits for the period 1850-1913 came to a margin of error between 10 to 15 per cent. Besides, for the analysis of long-term economic development a certain margin in the absolute figures is not too problematic. As long as the estimations methods are consistent, the relative developments of the various industries will be of sufficient reliability.

4. Developments in employment structures

As mentioned before, according to Kuznets an indicator of economic development is the agricultural share in employment and output. He found that as countries develop the share of the labour force working in the agricultural sector decreases. At first this is due to an increasing share employed in the industrial sector. In a later stage of development this standard economic theory predicts that the share of employment in the service sector starts to rise.

For Indonesia a number of data sources on the labour force is available. The most important are the population censuses held in 1930, 1961, 1971, 1980, 1990 and 2000. Furthermore intercensal surveys were conducted in 1976, 1985 and 1995. Moreover, since 1976/7 National Labour Force Surveys (Sakernas) are conducted annually with the exception of the years 1981, 1983 and 1984. For 1905 an enumeration is available, which is considered of poor quality in absolute terms. However, it gives a rather accurate picture of the relative distribution of the labour force. Together these data give a rather good picture of long-term developments in the employment structure. Figure 1 summarizes the findings on an aggregate level.

Figure 1: Occupational structure, 1905-2000

Sources: 1905: Jaarcijfers koloniën, 1909/10; 1930: Volkstelling 1930; 1960, 1971, 1980, 1990 and 2000: Sensus Penduduk. Note: The category ‘Others’ mainly consists of ‘activities not adequately defined’.