Political economy of land tax in Pakistan.

Qureshi, Sarfraz Khan

1. INTRODUCTION

Taxation of the agricultural sector is a major instrument for mobilization of the surplus to finance development projects within the agricultural sector and/or the rest of the economy. For many years, the need for a heavier taxation of agricultural land has formed part of the conventional wisdom regarding the ways of extracting agricultural surplus and increasing the tempo of agricultural development in poor countries. Land taxes have both equity and efficiency properties that gladden the hearts of both economists and vocal politicians belonging to urban areas. Taxes on land promote efficiency in the allocation of scarce resources by creating incentives for farmers to increase their effort and reduce their consumption, thus expanding the amount of agricultural produce available to the non-agricultural sectors of the economy. A tax on land has an important redistributive function because its incidence falls squarely on the landlord and is shifted neither forward to consumers nor backwards to suppliers of agricultural inputs; nor does it introduce distortions in the allocation of productive resources.

The conventional wisdom has had a difficult time in becoming conventional practice. The share of land taxes in total tax revenue is generally low in the developing countries and has been declining over time. In Pakistan, the share of land taxes in total taxes has fallen from 16 percent in 1960 to 6 percent in 1970, 2 percent in 1975 and, finally, just .5 percent in 1982. As a share of total agricultural income, land taxes have shown a similar downward trend. The decline stems in part from the fact that rates and assessments have not kept pace with the growth of agricultural incomes. Also, in some countries the taxes on land are either being abolished or rendered ineffective by exemptions and archaic assessment procedures. In India, some states have simply eliminated the land tax. In Pakistan, the land tax has recently been seriously diluted by the exemption accorded to irrigated land holdings below 12.5 acres and un-irrigated holdings below 25 acres.

The growing gap between theory and practice has its roots in the balance of political forces that determine tax policy in the developing countries. It is axiomatic that interest groups will attempt to shift the burden of taxation to others while reaping, to the maximum extent possible, the benefits of government expenditure. Since the agricultural sector is the dominant political force in many of the poorest developing countries, the erosion of land taxes in particular and of the taxes on rural incomes in general is almost an inevitable outcome in a world where voters and political leaders act on the basis of personal benefit-cost calculations, as has been eloquently pointed out by Anthony Downs [2]. In countries where rural interest groups predominate, the relevant question, then, is: How far can the agricultural sector shift the burden of taxation onto other groups?

The recent experience of Pakistan throws some interesting light on the political economy of the country's agricultural development because of the dramatic changes that the benefits of the government's development expenditures to the rural sector have undergone in the past two decades. In the early 1960s, yields began to rise as a result of the tubewell-supplied irrigation water and increased fertilizer applications, both heavily subsidized by the government. In the second half of the 1960s, these new inputs were joined by new seed varieties, whose multiplication and distribution were also the responsibility of the government. Under the circumstances, the farmers' resistance to land taxes or other forms of direct taxes should have weakened. But the share of taxes in Pakistan's GNP and the rural sector's share of those taxes have shown no tendency to rise. One explanation for this phenomenon, at least in the case of Pakistan, lies in another Downsian principle, namely that it is the perceived benefits and costs which are relevant to voters in deciding how they will cast their votes. From the point of view of the rural voter, a tax on his land is direct and real, while the benefits are diffuse, sometimes intangible, and frequently varying. If the perceived private returns to government expenditure could be raised, taxation of land might not only become politically agreeable but even be actively sought after by farmers as a means of raising their income.

2. THE SYSTEM OF LAND TAXES IN PAKISTAN

The present land tax system in Pakistan had its origins in the British tax system, introduced to the subcontinent in the 18th and 19th Centuries. (1) The British system was based on the classic economic principle that rent or the net output of land .(i.e. the difference between gross output and costs of the variable inputs) is an unearned source of income for landlords and can be taxed without any adverse effects on production. Over the centuries, the British developed an elaborate administrative machinery for administering the land tax system. Cadastral surveys were painstakingly conducted to measure land productivity. Data on crop prices and cultivation costs were compiled for different geographical sub-units. The assessment of land rental values (generally known in the subcontinent as settlement) was first made by the British in the second half of the 19th Century and have been periodically revised since then. A general reassessment was carried out in the 1920s and 1930s and special reassessments have subsequently been carried out in some districts. Except for a few agriculturally poor districts in the Baluchistan province, the Northern Areas and some of the former princely states, virtually all areas in Pakistan have been surveyed at least once for land tax purposes.

Two main systems of land tax assessment have evolved. The Punjab System is applicable to the provinces of the Punjab, the N.W.F.P., and most of Baluchistan. For the purposes of assessment, each district is treated separately. The district is further subdivided in clusters of much smaller areas for the purpose of assessment. In principle, the assessment area--or patwar circle, as it is commonly known--is supposed to be roughly homogeneous with respect to the general agricultural conditions, and an equitable assessment rate is fixed for all land lying within the area. The fields within the patwar circle are measured and classified according to soil type. The tax rate for each plot of land is then geared to the quality of the soil and expressed as a multiple of the assessment rate for the area. The basis of the assessment is the net produce accruing to the landowner. The gross value of output on each field is determined by multiplying together the yield per acre, the average acreage under each crop and the average price for each crop in the village market. The net produce (rental value) is obtained by deducting the share going to the tenant and the costs incurred in cultivation by the landowner from the gross value of production.

The assessed values remain fixed between the settlements that are generally made in each district at a 40-year interval. In the three districts of Lahore, Faisalabad and Sahiwal in the Punjab province, the assessed land values are reduced if the price of the main crop in the district for any year is lower than the price at which the physical output was valued during the year of the tax settlement. If the price is higher, the assessed values remain at the level fixed in the settlement.

The rate of tax is fixed at the time of tax settlement. This rate is applied to the rental value of cultivated land. Land left uncultivated does not incur any tax liability.

In areas other than the ones covered by the Punjab System, a slightly different method, known as the Sind System, is applied for fixing the assessed land values. This system is closer to the Ricardian doctrine of land taxation. The tax settlements are more frequent than in the Punjab System. For cotton, paddy and wheat, the rate of assessment per acre varies with crop prices in both the rising and falling phases of price changes. For other crops, the assessments remain fixed for the period between the two tax settlements. In theory, the Sind System is relatively more income-elastic than the Punjab System.

An outcome of the efforts by the British to assess land for taxation has been the legacy of an elaborate system of land tax assessment and collection and maintenance of records of land rights. The administrative advantages of this legacy for strengthening the land tax system in Pakistan are obvious.

3. ANOMALIES IN THE LAND TAX SYSTEM

The administration of the land tax system in Pakistan contains a number of anomalies, the first and most important of which is "the infrequent revision of the assessment rate. There has been no general revision in land assessments since the 1930s, although in some districts changes in the assessment rates are made annually. In Lahore, Faisalabad, and Sahiwal districts in the province of Punjab, assessment values are reduced if the price of the main crop in the district during the tax year is lower than the price of that crop in the year of the original tax assessment. This procedure is not symmetrical, however, and higher crop prices do not affect the assessment rate. In Sind, the tax administrators have attempted to revise tax assessments more frequently by taking into consideration the fluctuations over time in the prices of the principal crops grown in the province. Such efforts have proved difficult because landowners are a much greater powerful political force there than elsewhere in Pakistan.

A second anomaly of the system is that the rate and the tax base are fixed at the same time. A variable tax rate would be one way to compensate for annual fluctuations in crop prices. Nevertheless, the British fixed a statutory tax rate in the original legislation that established the land tax system and the tax rate cannot be changed without amending this legislation. Prior to 1871, the rate of the land tax was fixed at two-thirds of the assessed rental value. The Land Revenue Act of 1871 reduced this rate to 50 percent. In the 1920s, the tax rate was revised along with the assessed valuations, and the rate was reduced from 50 percent to 25 percent.

An additional anomaly in the procedures for establishing assessment rates is the legal ceiling on increases in the assessment rate for individual farms in the patwar circle. The Land Revenue Act of 1928 stipulated that the assessment could not be revised upwards by more than two-thirds for any single parcel of land or by more than one quarter for the assessment area as a whole. This limitation would not prove to be a substantial barrier if there are more frequent revisions in land assessments. But the tendency to revise land assessments at 40-year intervals and the secular increases in agricultural prices virtually ensure that the tax yield from the land tax will not rise as fast as increases in agricultural income in Pakistan.

Another anomaly is the narrow definition used of the net produce of land in establishing the assessment rate. No attempt is made to include the income generated by livestock or other on-farm activities. Income from these activities escapes all taxation, as agricultural income is exempted from the payment of income tax in Pakistan.

4. EQUITY ASPECTS OF THE LAND TAX

It is generally believed in Pakistan that land tax is a regressive and inequitable levy. It is argued that since tax settlements were carried out at different times in different areas, it is likely that wide variations in effective rates prevail. The horizontal and vertical inequities in the land tax are ascribed to the outdated assessed land values. Opponents to increases in land taxes have argued that as the designing and administration of a progressive land tax are a difficult task and the present system is inequitable, the burden of the land tax should not be increased by ad hoc increases in rates and/or assessed land values. In fact, the argument has often been advanced that since flagrant inequities are tolerated only when the tax is light, there is a need for lowering the rates of the land tax in Pakistan.

No empirical study on the incidence of land taxes has tested the hypothesis of this tax being a regressive one in Pakistan. The land tax system in Pakistan is badly in need of reform if it is to play any significant role in the generation of government revenue. Given the configuration of political forces in Pakistan, one important consideration in any reform of the land tax system will be its incidence. Based on the anomalies described above, it is impossible to say on an a priori basis whether the land tax system is progressive, regressive or neutral. A direct test of the progressivity of Pakistan's land tax system can be made by examining the share of taxes in the individual income for landowners of different income classes. However, in the absence of appropriate household-income data for landowning families, it is necessary to use more aggregative data. To determine tax progressivity, the share of land taxes in each district's income has been regressed against the per capita agricultural income of 37 districts of Pakistan.

The data needed for the analysis are: (1) tax collections by districts, (2) agricultural income by districts, and (3) agricultural labour force by districts. Before we present the results, a brief account of the data sources and their quality is given.

Tax Collection

Tax collections by districts are taken from the records of the Board of Revenue (the department which assesses and collects land taxes in Pakistan). District data are compiled by the office of revenue collectors from tehsil returns. The provincial data on land tax collections, as shown in various budget documents, are based on this set of data. The data on tax collections include collections under development cess, and are generally of high quality.

Agricultural Population

Since the data on agricultural income by districts were available for 1967-68, we projected the district rural population as recorded in 1961 census to 1967-68. The rates of growth used in this projection were based on the observed increase in population between the 1951 and 1961 censuses. The share of agricultural population in rural population in 1967-68 was based on the Population and Labour Force in Pakistan [3].

Gross and Net Values of Agricultural Production

The estimated district-wise agricultural income is based on an unpublished study conducted by the Planning Board in West Pakistan [4]. The data relate to the year 1967-68.

Agricultural income measured in each district is identical in concept to the "net value added by Agriculture" as conventionally estimated in national income accounting. It must be emphasized that the measured income refers exclusively to the income obtained from productive activity in agriculture, i.e. from crops, livestock, forestry, fruit and vegetables. No exact connection can be established between this income and the family or household income. Off-farm income, income from self-occupied houses and income transfers have not been taken into account.

The sources of data for production and prices of agricultural products are similar to the ones used in the estimation of national income originating in agriculture. The only difference is the use of district prices for valuation.

The inputs whose value was deducted included seeds, fertilizers, tubewell irrigation water, canal water and machine services. No input costs were deducted for livestock, as the corresponding fodder crops and pastures were not included in crop output. This may have resulted in some bias in the relative tax-paying capacities of the districts if the livestock were disproportionately distributed among them.

The inputs were valued at market prices that included indirect taxes and excluded subsidies in the valuation. This is in accordance with standard national income accounting procedures. This is also in consonance with the measurement of net income as an index of the tax-paying capacity.

The quality of data was reasonably good. The data on the gross value of production are expected to be more reliable than those on the net value of production. The reason for this is that the data on input use and input prices were not, as they are not even now, collected with the same reliability as the data on production and prices of agricultural produce.

As mentioned previously, the ratio of land taxes to district income was regressed on gross and net district agricultural incomes per agricultural labourer. The results are presented in Table 1.

Equations 1 and 2 relate to the 37 districts of Pakistan. The coefficients are small in size but highly significant and positive. The values of the elasticity of tax-income ratio (e) calculated at the mean value of district per capita income are higher than 2, which implies that the land tax is highly progressive as between districts. Given the cross-sectional nature of the data used in the regression analysis, the proportion of variance explained is quite high.

We tested the possibility of the progressivity found in the land tax being simply a statistical artifact. A scatter diagram of the percentage of income paid in taxes and district per capita income did not show any distinct clusters between high- and low-income districts. Nevertheless, we divided the districts into two categories: those above and those below the mean district per capita income. A separate equation was estimated for each category of districts. The results are presented in equations 3, 4, 5 and 6. The coefficients are significant and positive for each type of district. The elasticities are greater than one.

Separate equations for districts belonging to the Sind System and the Punjab System of land taxation were estimated. The results are presented in equations 7, 8, 9 and 10. The coefficients are positive and significant for both systems. The Punjab System is relatively more progressive than the Sind System. This is contradictive of the theoretical expectation. This finding may be explained by the nature of the administration of land taxes as well as by the different political power structures in the two regions. The Sind System may be administered in a way that large landowners pay somewhat lower taxes than are paid under the Punjab System. This probably explains Sind's reversion to the Sind System after the breakup of one unit in West Pakistan in 1968-69. In the new Provincial Assembly in Sind, the Sindhi landowners could and did legislate for reversion to the Sind System.

5. EFFICIENCY ASPECTS OF THE LAND TAX

Economists often stress the beneficial effects of increased land taxes on agricultural output, but the politicians generally argue that additional taxation adversely affects agricultural productivity. Given the level of government expenditure benefiting the agricultural sector, this section examines the effects of an increase in the burden of land tax on agricultural production in Pakistan. The cross-section data of different districts in Punjab for four years are analysed. We have regressed land tax per cultivated acre ([X.sub.1]) and three alternative proxy variables ([X.sub.2] or [X.sub.3] or [X.sub.4]) for government expenditure on the value of agricultural output per cultivated acre. The data sources for values of agricultural output, taxes and cultivated area were described in the previous section. No data are available on district-wise government expenditure on agriculture. We have used three proxy variables. The government has provided subsidized credit to farmers to enable them to buy fertilizer, install tubewells and purchase tractors and other items of agricultural machinery. Institutional credit in each district as a proportion of total institutional credit advanced to farmers in Punjab ([X.sub.2]) and institutional credit per acre in each district ([X.sub.3]) are proxy variables for government expenditure benefiting the agricultural sector. The relative acreage under high-yielding crop varieties is positively related to government expenditure on subsidized water, credit, fertilizer and, probably, agricultural machinery. Acreage under high-yielding varieties of wheat as a percentage of total wheat acreage in each district ([X.sub.4]) is a proxy variable for government expenditure in the years 1970-71 and 1972-73.

Table 2 presents the regression results. The land tax variable is positively and significantly related to the agricultural output per acre for all four years in all equations. Government expenditures on agricultural development are positively and significantly correlated with the agricultural output per acre for eight out of ten equations. For the two equations the variables are related positively but the government expenditure variable is statistically insignificant. It is thus safe to conclude that upward revisions in the level of land tax would help to increase agricultural productivity. This beneficial effect would be reinforced if the revenue from increased land taxes is spent on agricultural development through increased government expenditure benefiting the agricultural sector.

6. POLITICS OF THE LAND TAX

The archaic and inflexible land tax system in Pakistan can be traced to the political dominance of the landowning class who, almost as a matter of principle, have resisted any reforms in the land tax system. More than 40 percent of the voters in Pakistan own agricultural land but their representation in elected bodies has always been more than proportional to their numbers because of the influence which they exercise over tenants and landless agricultural labour. The political power of landlords was apparent in the earliest days of Pakistan and has not diminished markedly since then. (2) In the 1951 provincial elections in the Punjab, landlords won more than 80 percent of the seats, while in the 1955 Sind elections landlords claimed more than 90 percent of the seats. In 1971, the situation had not changed substantially. Out of 138 members of the West Pakistan Constituent Assembly, 105 were landlords. Although no analysis of the socio-economic background of the elected representatives in the National and Provincial Assemblies elected in 1985 is available, informed opinion holds that landlords have again been elected to the assemblies in a big way. Given this political complexion, it is not surprising that there has been no new legislation to reform the land tax system or to effectively introduce any formal agricultural income tax. (3)

The political influence of landlords is felt not only at the legislative stage but also in the administration of existing land tax laws. The collection of land taxes is the responsibility of the district administration and specifically of the District Commissioner, or the Collector, as he was sometimes known during the British period. The District Commissioner, however, is required by law to consult with members of the local community. This practice has provided considerable leverage to the local community in the area of land tax administration. Prior to the introduction of the system of Basic Democracies in 1962, local government took the form of a panchayat which invariably was made up of the representatives of the major landowning groups in each village. With the advent of the Basic Democracies system, the balance of political power shifted in form but not in substance. The chairmanship of the union council was generally rotated among the biggest landlords while the membership was made up of smaller landowners.

Landowners not only control administration of the land tax system through the political process but are themselves frequently the administrators. At the lowest level, the local official responsible for the collection of the land tax is the numberdar who is, in almost every case, a landlord owning the largest amount of land in a village. At higher levels, the tax officers at the district, divisional, provincial and federal levels are civil servants, the great majority of whom have ties with the landowning classes. As a result of the historic inequalities in the education system, the sons of landlords have had easier access to higher education than those of other rural classes, and were able to enter the civil service more easily than any other group. Similarly, the military, which at times has been a dominant political force in Pakistan, draws its officer corps mainly from the landed classes. Even though Pakistan was endowed at the time of its creation with a civil service in which rank was based on merit and not on social background, the fact that so many government officers had direct family ties with the landed classes has made effective reform and implementation of land tax laws difficult to achieve.

The tendency on the part of local tax officers to minimize tax liabilities on land can be seen from a recent case study of the operation of the land tax system in Multan district [1].

The settlement officer in 1966 underestimated the prices and yield per acre. The measurement of matured area in the circle was accurate. The costs of cultivation were underestimated. The underestimation of prices, costs and physical output cannot be explained because reliable data on these aspects are available from other government agencies. The yield data published by the government come from revenue officials who are also responsible for the settlement operations. The extent of underestimation is massive. The gross value of crop output in the 1966 settlement for the assessment circle was 12 million rupees. An alternative estimate based on more accurate price and yield data raises the value of gross output to 33 million rupees. Net income is similarly underestimated. The theoretical upper limit for land tax per matured acre was Rs 14.16 in the 1966 settlement. An alternative estimate based on realistic figures of yields and prices would have yielded the tax of Rs 31.09 per acre. The rate per matured acre actually fixed in the 1966 settlement was only Rs 3.64. This is much lower than the theoretical limit as calculated by the settlement officer himself. Interestingly enough, the actual assessment in 1966 was the same as that in 1921.

Two main factors explain the under-assessment of the land tax. The narrow and legalistic factor could have been the prescribed limits beyond which the assessment on the circle and village level cannot be increased between two consecutive settlements. This explanation is not relevant to the 1966 settlement in Multan, as the settlement officer did not propose any increase in the assessment over the 1921 level. It does not have much weight either, as a valid query can be made regarding the legal restrictions not having been nullified by legislation. The probable reason for the under-assessment is that either the government is weak and, therefore, hesitant to increase the tax on a whole mass of rural people to avoid any mass protest movement, or the government is being controlled by big landlords who do not want to tax themselves or their class.

7. SOME POLICY IMPLICATIONS

The central issue discussed in the paper is an explanation of the discrepancy between the conventional wisdom and conventional practice that land tax, despite being conceptually the most efficient and equitable way of raising resources from the agricultural sector, has been, and currently is, an insignificant source of tax revenue in Pakistan in actual practice.

We found some evidence that supports the conventional wisdom. The land tax system as designed and administered in Pakistan is, between districts, progressive with respect to both gross and net incomes. The land tax also has a favourable impact on agricultural productivity.

The main factor that explains the opposition to increasing land tax burden in Pakistan is the political power of the landowners, which they have so far used to block any effective method of land taxation. The finding that tax proceeds, when spent on strategic investment projects which promote agricultural development, stimulate agricultural productivity, suggests a way out of this difficult situation. If landowners are given a guarantee that proceeds from land taxes would be earmarked for and spent on agricultural development benefiting them in some direct way, the opposition to increased land taxes may vanish and make the use of increased taxes more feasible in development financing. In the case of Pakistan, district councils represent probably a suitable level of government for spending the earmarked funds obtained from increased level of land taxes. Once the political opposition fades away, the much-needed tax revenue can be collected by an ad hoc increase in land tax rates and/or assessed land values and an annual readjustment of these by indexing them to an index of the prices received by farmers.

The long legacy of the British in the administration of this tax is an important positive factor in its use as a revenue-raising device. The legal limits on the increases in assessed land values between two consecutive settlements can easily be struck off the statute books. Given the progression in the present land tax system, there is no special need for exemptions and graduated rates, as such devices make the structure of the land tax too complex for its effective administration.

Comments on "Political Economy of Land Tax in Pakistan"

I am extremely grateful to the organizers of this Conference for inviting me to be a discussant of Dr Qureshi's paper. I must congratulate Dr Qureshi on his efforts to examine rather sensitive aspects of a tax, which until a few decades ago was a major revenue source of the provincial governments.

Briefly stated, Dr Qureshi's paper follows the historical assumptions that the land tax induces agricultural productivity and redistribution of wealth and income; it has both equity and efficiency properties; it creates incentives to increase farmers' efforts and reduces their consumption; and it is neither shifted forward nor backwards, nor does it introduce distortions in the allocation of productive resources. The author has briefly described the system of land taxes followed in Pakistan and has referred to some of its anomalies--on both policy and administrative considerations. As against the general belief that land taxes in Pakistan are regressive and inequitable, the author has made an attempt, based on an empirical study of the data pertaining to the year 1967-68, to establish that the land tax is highly progressive as between districts. The methodology followed by the author is one of regressing the ratio of land tax to cross-district gross/net per capita agriculture income of agricultural labour.

From the efficiency viewpoint, the author has not come out with any clear findings--the limitation being, of course, the non-availability of the appropriate economic data. The author has, nevertheless, stated that the tax has a favourable impact on agricultural productivity.

My comments on Dr Qureshi's paper relate to both policy and administrative issues. However, before I take up various aspects of the paper, let me submit that since the introduction of the Islamic levy Ushr in 1983, the land taxes stand abolished to the extent of approximately 97 percent of the total levy. The balance of 3 percent is payable only by non-Muslims and 'non-Sunni' sects of Muslims. As such, in Pakistan's context, we are here discussing the theoretical and administrative aspects of a levy which for all practical purposes is non-existent. Hence, our discussion is merely a theoretical exercise. Let me now address myself to Dr Qureshi's paper. My comments are as follows.

With regard to the progressivity of land taxes, Dr Qureshi's observations suffer from many constraints. Firstly, the land tax, being a flat tax based on the "land settlement" of each parcel of land, can at best be termed a 'proportionate tax'. The question of its being progressive or regressive does not arise. Secondly, Dr Qureshi's findings are based on regression of cross-district data--which in theory compares different size distributions of land and different income distribution. The better alternative would have been to regress the data on different farm sizes of the same quality of land. Under these objective situations, the comparison would have been more realistic. As such, his contention that land tax in Pakistan is progressive is not fully proved.

With regard to the equity aspects of land tax, the author has, perhaps, based his findings on the definition of equity as "increase in tax with the increase in income". In the given situation, his argument that land tax is equitable does not seem to hold good in so far as that the land tax is charged on a flat rate on the value of land. Given that one set of parcels of land is different in quality from the other in each district and is capable of yielding low or high income and that the land tax is charged on land rather than on income, the increase in taxes with increase in income, perhaps, is not a true indicator of the equity of land tax.

As to the legislative and administrative aspects, I do not feel very happy with the existing system. As rightly pointed out by Dr Qureshi, the produce indices as the tax base for land taxes is not a frequent exercise. Forty years' periodicity, as in the Punjab, and fifteen years', as in Sind, is appalling. The land structure, soil type, exogenous factors like the replacement of perennial irrigation system by all-season irrigation system, quality fertilizer and pesticides--all have dramatically changed the situation. Regular 'settlement' of land with a frequency of, say, 5-10 years perhaps, is the only way of producing a realistic tax base. The administrative anomalies are, of course, in terms of appropriate 'settlement of land'. The present mode of 'settlement' essentially depends on both the historical considerations and the value judgement of the government functionaries, who are in no way experts in soil analysis, and the reasonably expected output from a parcel of land. The unseen inputs from numberdars, who themselves are the landowners of the area, further aggravate the 'settlement' process.

Thus, the taxation of agricultural income in Pakistan is essentially a question of public policy, which happens to be predominantly influenced by the landed aristocracy. The agriculture sector lobby, being the most influential, vocal, and represented in politics, has decided against the proposed levy! The pragmatic approach would, therefore, be to move slowly and step by step--first, perhaps by introducing tax on income of orchards. Out of the 214937 farms reporting orchards (orchard area being 235042 hectares), a total of 234178 acres belong to the private sector. The cropped area above 10 hectares, being 120270 hectares, could perhaps be the ideal orchard size for tax purposes as an initial step.

Ahmad Khan

Assistant Commissioner of Income Tax, Multan

REFERENCES

[1.] Azhar, B. A. "Land Revenue Assessment: A Case Study". Pakistan Development Review. Vol. XXII, No. 3. Autumn 1982.

[2.] Downs, Anthony. Economic Theory of Democracy. London: Harper and Row Ltd. 1957.

[3.] Pakistan. Central Statistical Office. Population and Labour Force in Pakistan, 1967-68. Karachi. 1977.

[4.] West Pakistan. Planning Board. "Agricultural Income in West Pakistan, 1967-68". Lahore. (Mimeographed)

(1) For a concise description of the history of land taxes in Pakistan, see Chapter 8 in Mahmood Hasan Khan's book, Underdevelopment and Agrarian Structure in Pakistan (Boulder, Colorado: Westview Press. 1981).

(2) For an interesting analysis of the political importance of the landed class, see Shahid Alam, "Economics of the Landed Interests--A Case Study of Pakistan". Pakistan Economic and Social Review. Spring 1974.

(3) For a view that land reforms have been hindered by these same forces, see R. Herring and M. Ghaffar Chaudhry, "The 1972 Land Reforms in Pakistan and their Economic Implications: A Preliminary Analysis". Pakistan Development Review. Vol. XIII, No. 3. Autumn 1974.

SARFRAZ KHAN QURESHI *

* The author is Chief of Research at the Pakistan Institute of Development Economics. He is thankful to Professor Stephen E. Guisinger, Dr Sohail J. Malik and Dr M. Irfan for their valuable comments on an earlier draft of the paper. The author alone is, however, responsible for any errors and omissions.

Table 1

Regressions of Percentage of Income Paid in Taxes
on per capita Income

Equation Sample Estimated Equation
 No.

 1 37 Districts T/[Y.sub.g] = -.01218 + .00002 [Y.sub.g]
 (4.43001)

 2 37 Districts T/[Y.sub.n] = -.01365 + .00003 [Y.sub.n]
 (5.17161)

 3 19 Below Mean
 Income
 Districts T/[Y.sub.g] = -.00142 + .00001 [Y.sub.g]
 (1.912)
 4 19 Below Mean
 Income
 Districts T/[Y.sub.n] = -.00453 + .00001 [Y.sub.n]
 (2.495)

 5 18 Above Mean
 Income
 Districts T/[Y.sub.g] = -.01509 + .00002 [Y.sub.g]
 (1.811)

 6 18 Above Mean
 Income T/[Y.sub.n] = -.0177 + .00003 [Y.sub.n]
 Districts
 (2.244)

 7 12 Districts
 under
 Sind System T/[Y.sub.g] = -.00283 + .00002 [Y.sub.g]
 (2.128)

 8 12 Districts
 under
 Sind System T/[Y.sub.n] = -.00211 + .00002 [Y.sub.n]
 (2.068)

 9 25 Districts
 under
 Punjab System T/[Y.sub.g] = -.00020 + .00001 [Y.sub.g]
 (2.686)

 10 25 Districts
 under
 Punjab System T/[Y.sub.n] = .00024 + .00001 [Y.sub.n]
 (2.637)

Equation Sample [[bar.R] e
 No. .sup.2]

 1 37 Districts .36 2.039

 2 37 Districts .44 2.22

 3 19 Below Mean
 Income
 Districts .18 1.48

 4 19 Below Mean
 Income
 Districts .27 1.05

 5 18 Above Mean
 Income
 Districts .17 1.63

 6 18 Above Mean
 Income .24 1.81
 Districts

 7 12 Districts
 under
 Sind System .31 1.61

 8 12 Districts
 under
 Sind System .30 0.88

 9 25 Districts
 under
 Punjab System .24 1.72

 10 25 Districts
 under
 Punjab System .23 2.946

Notes: (1.) T, [Y.sub.g] and [Y.sub.n] are respectively per
capita tax, per capita gross income and per capita net income.

(2.) Figures in parentheses are t-values of the coefficients.

Table 2

Estimates of Regression of Value of Agricultural Output per Acre on
Land Tax per Acre and Indices of Government Expenditure

 Variables

Equation Years Intercept [X.sub.1] [X.sub.2]
 No.

 1. 1958-59 70.940 +9.086 +1.76
 (5.401) (2.695)

 2. 1958-59 17.348 +9.73
 (5.413)

 3. 1959-60 87.440 +10.05 +1.78
 (4.180) (2.050)

 4. 1959-60 89.018 +10.75
 (4.312)

 5. 1970-71 85.427 +29.10 +6.71
 (3.120)

 6. 1970-71 84.684 +32.123
 (4.655)

 7. 1970-71 64.144 +19.81
 (4.250)

 8. 1972-73 140.700 +55.93 +8.53
 (5.819) (3.314)

 9. 1972-73 126.265 +61.81
 (6.403)

 10. 1972-73 102.466 +39.35
 (5.203)

 Variables

Equation Years [X.sub.3] [X.sub.4]
 No.

 1. 1958-59

 2. 1958-59 +0.56
 (1.991)

 3. 1959-60

 4. 1959-60 +0.41
 (1.433)
 5. 1970-71

 6. 1970-71 +2.54
 (2.272)
 7. 1970-71 +1.93
 (6.076)
 8. 1972-73

 9. 1972-73 +3.94
 (2.452)
 10. 1972-73 +2.78
 (5.546)

Equation Degrees of [[bar.
 No. Years Freedom R].sup.2]

 1. 1958-59 19 0.72

 2. 1958-59 19 0.67

 3. 1959-60 19 0.61

 4. 1959-60 19 0.59

 5. 1970-71 19 0.67

 6. 1970-71 19 0.61

 7. 1970-71 19 0.65

 8. 1972-73 19 0.74

 9. 1972-73 19 0.74

 10. 1972-73 19 0.94

Notes: (a) Figures in parenthesis are t-values of the coefficients.

(b) [X.sub.1] = Land tax per acre (in rupees).

(c) [X.sub.2] = Institutional credit in each district as a proportion
of total institutional credit advanced to farmers in Punjab.

(d) [X.sub.3] = Institutional credit per acre (in rupees).

(e) [X.sub.4] = Acreage under high-yielding varieties of wheat as
percent of total wheat acreage.