AN ABSTRACT OF THE THESIS OF RACHED AKROUT for the degree of MASTER OF SCIENCE in Agricultural and Resource Economics presented on /0-otO "?& Title: ECONOMIC ANALYSIS OF THE LIMITATIONS TO THE PRODUCTION OF BREAD WHEAT IN TUNISIA Abstract approved: // Q John A. Edwards In order to effectively use wheat prices as tools of government policy, it is necessary to investigate the possible effects of changes in these prices on different economic variables. In this study, at- tempts have been made to estimate the effects of changes in wheat prices upon: 1. The allocation of the available wheat acreage between bread wheat and durum wheat; and 2. The use of fertilizer for further improvement of wheat yields. In studying the farmer's acreage and yield response, respectively special attention was given to economically controlled factors included in the models, and estimates of the responses were developed for two historical periods - 1949-1963 (the traditional period) and 1960-1974 (the modern period). A detailed analysis of the behavior of economic factors and a comparison of their magnitudes between the two major periods of production are presented. The results of the empirical study of farmer's acreage supply response and yields response indicate that: 1. Tunisian farmers are significantly responsive, in their acreage allocation decisions, to relative output prices; 2. Yields of the new high potential bread wheat varieties, recently introduced in Tunisia, are more responsive to fertilizer use, under dry-land conditions, than the local varieties; and 3. The demand for fertilizer by Tunisian farmers is significantly determined by farm-income and relative output prices. This suggests that a policy which would raise the price of bread wheat relative to that of durum wheat, other things the same, would contribute to increasing overall bread wheat production and therefore reducing governmental spending on bread wheat imports. Economic Analysis of the Limitations to the Production of Bread Wheat in Tunisia by Rached Akrout A THESIS submitted to Oregon State University in partial fulfillment of the requirements for the degree of Master of Science June 1976 APPROVED: Professor J/tmn A. Edwards in charge of major ,.Y " ' Head of Department dlfl Agricultural and Resources Economics Dean of Graduate Schoo' Date thesis is presented Typed by Susie Kozlik for /0~&l&~7<^ Rached Akrout ACKNOWLEDGMENTS The author owes a special debt of gratitude to Dr. John A. Edwards, major professor, for his ideas and valuable suggestions at all stages of this study and guidance throughout the author's graduate program at Oregon State University. He would like to thank Drs. Bruce R. Rettig, Roger G. Petersen and Timothy M. Hammonds, members of his graduate committee. The author is under a deep obligation to his colleagues in the Bureau du Plan et du Developpement Agricole, Tunis for collecting and arranging the data which made this research possible and encouragement throughout the entire period of his graduate study. Special thanks are also due to Drs. Malcolm J. Purvis and Reynold P. Dahl -- the Department of Agricultural and Applied Economics --at the University of Minnesota, for their help at the initial stage of the study. Indebtedness to my family for their encouragement throughout this study, should be particularly mentioned. TABLE OF CONTENTS Chapter I Page INTRODUCTION Problem Previous Related Researches Objectives II METHODOLOGY Hypotheses Methodological Basis The Acreage Supply Response Yield Response Estimation Procedures Acreage Supply Response Nerlove-Type Acreage Model The Yield Model III IV EMPIRICAL RESULTS 1 1 4 6 8 8 9 9 13 14 14 14 15 18 The Acreage Supply Response The Traditional Production Period The Modern Production Period Yield Response The Traditional Production Period The Modern Production Period The Demand for Fertilizer The Traditional Production Period The Modern Production Period 18 18 19 25 25 27 29 29 31 INTERPRETATION OF THE ESTIMATED RESULTS 33 The Acreage Supply Response The Traditional Production Period The Modern Production Period The Yield Response The Traditional Production Period The Modern Production Period The Demand for Fertilizer The Traditional Production Period The Modern Production Period Comparison 33 33 37 40 40 42 42 42 43 44 Chapter V Page POLICY IMPLICATIONS The Acreage of Bread Wheat The Yields of Bread Wheat BIBLIOGRAPHY 45 46 51 60 APPENDIC ES Appendix A: Basic Data Appendix B: Estimation of Acreage and Yield Responses for Durum Wheat 63 72 LIST OF TABLES Table 1 2 Page Tunisian domestic production and imports of bread wheat during the period (1949-1974), 3 The estimated acreage supply response in the traditional production period 20 The estimated acreage supply response in the modern production period 23 The estimated yield response in the traditional production period £6 The estimated yields response in the modern production period 28 The estimated demand for fertilizer in the traditional production period 30 The estimated demand for fertilizer in the modern production period 32 The'revised estimated acreage supply response in the traditional production period 35 The revised estimated acreage supply response in the.;modern production period 38 The acreage of wheat on Tunisian and European farms during the period (1949-1958) 41 11 The average acreage of wheat in Tunisia (1949-19'74) 48 12 The Tunisian exports of Durum wheat (1957-1966) 49 13 The estimated per capita consumption of wheat bydegree of urbanization in Tunisia (1966) (in kilograms) 51 14 Projection of the production of cereals in Tunisia 52 15 Acreage adjustment to changes in relative prices (a) 5?'' 16 Acreage adjustment to changes in relative prifces (b) 58 3 4 5 6 7 8 9 10 LIST OF APPENDIX TABLES Table 1 Page The acreage and yields of bread wheat and Durum wheat in Tunisia (1949-1974) 64 The yields of bread wheat and the total annual rainfall in Tunisia (1949-1974) 65 The wholesale prices of bread wheat, Durum wheat and nitrogen in Tunisia (1949-1974) 66 The yields and prices of bread wheat relative to the yield and price of Durum wheat and to the price of fertilizer (1950-1974) 68 5 The estimated per capita income in Tunisia (1949-1974) 69 6 Annual yields and per hectare nitrogen use in bread wheat production during the period (1949-1974) 70 The estimated Durum wheat acreage supply response in the traditional production period 75 The estimated Durum wheat acreage supply response in the modern production period 77 The estimated Durum wheat yield response in the traditional production period 79 The estimated Durum wheat yield response in the modern production period 81 2 3 4 7 8 9 10 LIST OF FIGURES Figure 1 2 Page Acreage of bread wheat and Durum wheat in Tunisia (1949-1974) 47 Bread wheat acreage adjustment to relative price changes 56 ECONOMIC ANALYSIS OF THE LIMITATIONS TO THE PRODUCTION OF BREAD WHEAT IN TUNISIA I. INTRODUCTION Problem The production of wheat in Tunisia has been a dominant traditional agricultural activity for centuries. It started with the Romans who introduced the first seeds from different areas of their Empire and continued through the recent French protectorate period during which new techniques and processes of production were implemented in the area. In recent years, more than a third of the 3. 4 million hectares of the cultivated land in the country have been devoted to wheat production with 1/5 or less in bread wheat. However, the major prob- lem that characterized the production was, and still is, the presence of a high variability in production levels from one harvesting year to the other resulting in significant importation trends, specifically of bread wheat, in the country. As a matter of fact, the importation of bread wheat in Tunisia was a recognized tradition since the introduction of this agricultural commodity by the French in the nineteenth century, but in the last decade, it has taken a significant place among the imported agricultural commodities in the country since bread wheat imports accounted, on the average, for more than 70% of the total imported agricultural commodities. In an average year like 1962/1963 for example, the value of imported live animals, dairy products, fruits and vegetables (which represent the major imported agricultural commodities in Tunisia) was 8.2 million U. S. dollars, contrasted with 19 million U. S. dollars of imported bread wheat. On the other hand, the percentage of imported bread wheat to the domestic total production reached extremely high levels during the last 15 years. In fact, as is illustrated in Table 1, the average of imported bread wheat as a percentage of the local production in the traditional production period (1949-1963) was 118. 3, contrasted to an average of 256. 6 during the modern production period (1960-1974). This high importation trend of bread wheat is very suggestive, in that, even with the recent transition from the past traditional production process to a relatively modern agriculture, the reliance of the domestic demand on imported bread wheat is even higher. In addition, with the recent emergence of a large tourist industry in the country (which counts for about one fourth of the domestic population), and the high domestic population growth (3%), future bread wheat production stimulations are inevitable and certainly needed to meet the increasing domestic consumption trends. The present research covers a very definite period (i. e. , 19491974) during which the production of bread wheat was characterized by The information concerning bread wheat imports and values was gathered from different issues of the United Nations trade yearbook. Table 1. Tunisian domestic production and imports of bread wheat during the period (1949-1974). Total Domestic Production of Bread Wheat (1000. tons) 1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 197^ 1973 1974 Source: 180 180 120 230 200 190 100, 150 140 130 120 90 43 93 130 95 120 70 75 73 80 150 200 258 235 202 Total Tunisian Imports of Bread Wheat (1000 tons) 14. 9 69. 4 4. 0 39.9 9.8 0;6 18.0 130. 1 118. 1 26.8 68.0 154. 1 366.5 271.8 157. 6 189.6 141. 3 249. 7 338. 7 211.6 381. 3 249. 1 16Y\0 212.0 291. 6 127.0 Tunisian Bread Wheat Imports . nn r ^ x 100 Domestic Prod[uction 8.0 38.5 3.0 17. 3 4.9 0.0 18.0 86.0 84.0 20.6 56.6 .171.2 853.0 292.2 121.2 199.5 117. 7 356.7 451. 6 289.8 476. 6 166.0 83.5 82. 1 124. 1 62.8 Ministere de I'agr icultur e, biireaudu plan et du development agricole, enquetes statistiques. two distinct stages: 1. A first traditional production period which covers the range (1949-1963), and during which the production of bread wheat was highly extensive, and essentially reliant on economically uncontrolled factors; 2. And a modern production period covering the years (19631974) during which a new high yield technology was introduced in the production process essentially through the new bread wheat varieties, and economically controlled factors were sought for the improvement of domestic production levels. In addition, under each of these two major stages, the aggregate bread wheat production in the country was characterized by: A - A high variability of the annual average yields associated with B - A relatively significant variability of the acreage devoted by the Tunisian wheat producers to bread wheat production from one harvesting year to the other. Previous Related Researches In the available research studies related to this vital problem, the work of MM John D. Hyslop, Reynold P. Dahl, and Malcolm J. Purvis, 2 is acknowledged and highly contributive. The former two researchers, in an economic report submitted to the Tunisian Department of Agriculture in 1970 (10), tried to investigate some of the facts that were, to an extent, at the origin of the wheat acreage variability 2 MM John D. Hyslop, Reynold P. Dahl and Malcolm J. Purvis are Agricultural Economists, past members of the University of Minnesota team in Tunisia. in Tunisia on the basis of historical data covering the period 1950- "■I960. Among their findings, the influence of world wheat prices as well as climatic conditions in the area are worthwhile mentioning. In addition, in the same report, it was mentioned that the high variability of domestic wheat yields during the decade 1950-1960 w ere dependent upon the following three factors: , : 1. The rainfall in the area; 2. The changes in the cultural practices by the Tunisian farmers; 3. And some other political events. Even though, the findings are most provocative, there was no attempt, however, to quantify them for a further investigation of the problem. On the other hand, M. Malcolm J. Purvis, in a recent article entitled "The New Varieties Under Dry-land Conditions: Mexican Wheats in Tunisia, " and published in the American Journal of Agricultural Economics (February 1973), tried to give some thoughts to the problems facing the introduction of the Mexican varieties in Tunisia. Specifi- cally his main concern was to compare the yields of the local varieties and those of the Mexican high yielding varieties under dry-land conditions. And even though his findings are very well founded, there is still some doubt about the degree of response of the local and the new varieties to fertilizer, under similar dryland conditions. Objectives At the international level, an ever growing chorus of economists and politicians warns of an impending, if not already existing, crisis in world agricultural production and specifically in developing countries. The President of the International Bank for Reconstruction and Development, Georges D. Woods, has stressed the necessity for "much greater emphasis on increasing production on the land both to feed growing populations and to feed growing industries" (4)., T. W. Schultz, has addressed a book to this topic (20). In Tunisia, particularly, a similar urgency for increasing the production of bread wheat to meet the high domestic consumption has been recognized; an increasing concern of the national agricultural agencies as well as the Tunisian producers with the production of bread wheat has developed and wide-range extension programs have been undertaken as a result. In contrast to this rapidly growing consensus about the existence of critical problems in world agricultural production, and specifically of bread wheat production in Tunisia, little agreement exists over what policies might improve future prospects. Jere B. Behrman (2) is quoted to say In the professional economic literature and in the policy discussions about underdeveloped agriculture, considerable debate has raged over.the merits of various proposals. Concurrence has not been possible, however, partly because conflicting policy recommendations are implied by the various .proposed "a priori" Hypothesis "about the supply responsiveness of underdeveloped agriculture. The present study is intended to give some insight into the response of the Tunisian bread wheat producers to economic factors in their acreage allocation decisions, on one hand, and the response of the yields to economic factors on the other. Specifically, the study intends: 1. to analyze the acreage supply response to economically controlled and uncontrolled factors; 2. to analyze yield response to economically controlled and uncontrolled factors; 3. and to give some thought to the possibilities and means of increasing the domestic production of bread wheat under the existing economically uncontrolled limitations. II. METHODOLOGY Hypotheses In the view of any economist, prices are important determinants of economic behavior. As a matter of fact, in his introductory eco- nomic analysis (18), Paul A. Samuelson wrote: At a higher price of wheat, farmers will take acreage out of corn cultivation and put it into wheat, in addition, each can now afford the cost of more fertilizer, more labor, more machinery, and can even afford to grow extra wheat on poorer land. All this tends to increase output at higher prices. However, it was argued by many western agricultural economists (2) that in underdeveloped economies (Tunisia is one of them), the response of farmers to changes in the prices of their output, in their acreage allocation decisions, is irrational and uneconomical because most of these countries are experiencing subsistence economies. Also, it was argued, as mentioned earlier, that the new high yielding varieties, introduced in Tunisia during the modern production period, were developed under irrigated conditions, and their response to fertilizer applications under dry land conditions such as the ones existing in Tunisia, is inferior to the local varieties. The main hypotheses to be tested here are: 1. That the Tunisian farmers are responsive to the relative prices of bread wheat to Durum wheat, among others, -* ^The other factors, however, will be explicitly discussed in the following chapters of this study. in their acreage allocation decisions; 2. And the new wheat varieties introduced recently in the Tunisian agriculture, are more responsive to fertilizer applications (among others) than the local variety. And, as a result of the second hypothesis, we shall be concerned also with a secondary hypothesis; that the demand for fertilizer by the Tunisian farmers is responsive to the farm income and the relative prices of bread wheat to fertilizer. Methodological Basis The Acreage Supply Response In the American economic literature at least, supply response has been for more than two decades a central economic issue. In fact, one of the earliest attempts to apply this concept to production stimulation of certain American agriculture products was undertaken by W. B. Hubbard in 1923 (7). In 1925, B. Bradford Smith (22), noticed that besides the existence of an economic psychological relation between price and subsequent acreage, there is a need for a more intensive analysis in which the selling price of the selected crop relative to the possible crops should be considered, because this relative price is the chief factor 4 The other factors, however, will be explicitly discussed in a later chapter. 10 influencing the net returns of the farmer. In 1929, a leading agricul- tural economist, J. D. Black (3), complained that many factors other than the preceding year's relative price should be considered in any investigation of acreage variations, such as damage from pests and diseases, improvements in machinery for handling competing crops, improvements in varieties or adaptability of competing crops. The same year, in a study of farmer's response to price, L. H. / Bean (1) noticed that the dominant factor in the change of acreage in a given year for four different areas and nine commodities was the preceding output price received by the farmers. In addition, he pointed out that factors such as the yields, weather, and profits have to be taken into consideration because they may be assumed to influence the , farmer's actions, Recently in a study of the supply dynamics of cer- tain agricultural commodities in the United States, M. Marc Nerlove advanced the theory that: 1. the observed acreage devoted to an agricultural commodity —' in any given year was linearily related to the acreage of the preceding year adjusted to the difference between the expected long-run equilibrium acreage (i. e. , the acreage which would be reached if all the necessary time taken by the farmers for price change adjustments _ was available) and the preceding year observed acreage; 2. the expected price for the farmer's output in any given year is linearily related to the preceding year expected price of output 1.1 adjusted to the difference between the preceding year observed and expected prices. By a simple process of substitution, he obtained a general linear relationship between the actual acreage, the lagged observed price of output and the lagged observed acreages (for one and two periods). In other words, the Nerlovian supply dynamics system is expressed as the following: (1) X^X^ + YfX*-^) >< * (2) P^P^ + PfP^-P^) (3) X t = % + a l ^ where V and Pare the coefficients of adjustment and expectation respectively and such that: o < Y < 1; O<(3<1 X is the desired long-run equilibrium acreage'. P is the expected price of output for a given harvesting year. a, is a positive constant. and t refers to time. Substituting (3) and (2) into (1), we have: X^X^ + v^+a^-X^) X t = X t-1 + ^o + *lt Pt-1 + P (P t-l " ^t-l^ " YX trl 12 X Xt - Yao + (l.v, X^j + Xt=vao + + pP Xt = Yao ( Ya1 (1-P) (l-Y)Xt_1+va1 (l-P) - a ^^ <Xt-l - 0) X + PP,.! t-2 + ^Xt-2 - Vao> t-1 + (1-Y) Xt_1 + (1^)(Y-1) Xt_2 + (l-P) Xt_1 - Y- (l-P) + YajpP^j Xt = Yao - Yao (l-P) +[ (l-P) + (1-Y)] X^ - (l-p)(l-Y) ^t_2 + va PP Y 1 H t-1 Xt = aoYp+ [(l-P) + (1-Y)] Xttl - (1-P)(1-Y) Xt_2 + yaj PP^j And if we designate: (4) ,o = ao YP (5) uj = (l-P) + (1-Y)(>°) (6) ^ = (7) -(1-Y)(1-P)(<O) ^3 = YajP (>o) we can express the Nerlovian acreage supply model as the following: X = if + / X +u0X 0 + iT0P ,+u7 t o 1 t-1 2 t-2 3 t-1 t 13 Yield Response In a recent study covering the period 1944-1962 and concerning the dependence of the yields of bread wheat in Tunisia on the annual rainfall, an USAID Meteorologist, M. Lee Dutcher, found that during dry years, the yields of bread wheat were concentrated below the period average yield, and during the humid years, the yields were concentrated above it. The correlation was^ significantly high. In addition, he noticed that the correlation is even higher if the rainfall between September and April 5 instead of the total annual rainfall was taken into account. On the other hand, in their economic report (Production de ble en Tmnisie, tendences et variabilites), MM. John D. Hyslop and Reynold P. Dahl mentioned the impact of the changes in the rotation of crops on the yields of wheat in Tunisia after the Second World "War. M. Malcolm J. Purvis (16), in his discussion of the green revolution in the developing countries and in Tunisia particularly, noticed that the new varieties of soft wheat are very responsive to fertilizer application. 5 i These results were taken from "Production de ble en Tunisie, tendences et variabilites" by John D. Hyslop and Reynold P. Dahl, May, 1970. 14 Estimation Procedures Acreage Supply Response Nerlove-Type Acreage Model On the basis of the Nerlovian general supply dynamics theory, the actual acreage devoted each year to bread wheat production by the Tunisian farmers (X.) is hypothesized to be partly related to the lagged observed relative prices of bread wheat to Duram wheat, and to the lagged observed acreages for one and two periods, such that: X t and tr 3 = Tr o + 1T l X t-1 + ^Xt-2 + "Tz[^jt_1 + utwhere V *2 are constants, —— is the relative price of bread wheat to P D Durum wheat and u an error term which could partly be explained by the relative lagged yields of bread wheat to Durum wheat and the total annual rainfall, as it was pointed out earlier in the pioneering work of MM, B. B. Smith, J. D. Black and L. H. Bean. In other words, the Tunisian farmers' acreage allocation decisions are hypothesized to be partly influenced by output relative prices, relative yields, the preceding years acreage allocations, and the total annual rainfall in the area. Explicitly, the full acreage supply model would take the form: 15 + where: R "5 t + U t X = the actual acreage of bread wheat ——I = the lagged relative prices of bread wheat to Durum D/t-l wheat X = one period lagged acreage of bread wheat X = two periods lagged acreage of bread wheat , D/t-l the lagged relative yields of bread wheat to Durum wheat R = the actual total annual rainfall U = random error term reflecting unidentified factors a , a ... a = average supply response parameters to be estimated. The Yield Model On the basis of the research already mentioned, the yields of bread wheat in Tunisia are hypothesized to be linearily related to fertilizer application, the total annual rainfall and the acreage devoted to this agricultural commodity. 16 Explicitly, the statistical relation is: 6 Y = b + b (F ) + b'tR) + b. (A ) +i t o It 2t 3t t where: Y = the average annual yield of bread wheat; F = the amounts of fertilizer applied per hectare; R = the total annual rainfall; A = the acreage of bread wheat; I = random error reflecting unidentified factors; b ,b o 1 , b0 = yield response parameters to be estimated. 3 In addition, an attempt is made in this study to estimate the demand for fertilizer by Tunisian farmers as a response to changes in the prices of their output. Economic theory suggests that the quantity of a given fertilizer demanded depends on the price of that fertilizer, the prices of related commodities, and income (6). This could be written as the following: A more complete yield response model was developed here, such that: y t = K + (a - a F )F + (6 - B R )R + yRF + e oltt oltt ttt or: yt = K+«Ft+aFt+pRt+bRt +gV"t+et where: y, F, and R are bread wheat yields, fertilizer and rainfall respectively; K, a, a, (3, b and g are yield response parameters; However, the estimated coefficients associated with the square and cross-product terms were insignificant in the modern period. 17 F =C +C t o 1 ^RV * P_ F t 1 + CM +vY 2 t-1 t where! F = the quantity of fertilizer demanded by Tunisian farmers for bread wheat production; (P ) = the lagged price of bread wheat; B t- 1 F F = the actual price of fertlizer paid by Tunisian farmers; t M "Y = the lagged income of Tunisian farmers; = random error reflecting unidentified factors; C , C , C_ = fertilizer demand parameters to be estimated, o i 2 18 III. EMPIRICAL RESULTS The Acreage Supply Response The Traditional Production Period The acreage response model was fitted to time series data covering the period (1949-1963). 1. The data used are: The wholesale prices of bread wheat relative to that of Durum wheat expressed in millimes .7 per quintal, which are re- ported by the appropriate service of the Tunisian Department of Agriculture at the end of each harvesting year; 2. The total annual rainfall as the national total annual rainfall in millimeters, which is reported by the world weather report at the 6nd of each year; 3. The acreage of bread wheat harvested each year, in 1000 hectares, and reported either by the'Department of Agriculture or the Department of National Economy; 4. The national annual average yields, computed as the ratio of the total annual production to the acreage harvested in the country. They are reported by the appropriate service in the Department of Agriculture in quintaux per hectare. 7 Millime is a Tunisian currency (1 dinar = 1000 millimes *s 2. 75 U. S. dollars). 19 The number of observations is 13. Five regression equations have been estimated using a stepwise statistical procedure. The estimated acreage supply response functions are summarized in the following table. The estimated coefficient of the relative price of bread wheat to Durum wheat is positive, highly significant in regression (1), and significant at the 80% level in regressions (Z) and (3). They have relatively small standard errors and the R's squared associated with the regressions (1) and (2) are statistically significant. However, the estimated coefficients of the total annual rainfall, the one period lagged acreage, and the lagged relative yields, even though positive, are not statistically significant, their standard errors relatively large, and the R's squared associated with these variables are poor. The coefficient of the two periods lagged acreage is negative but statistically insignificant. On the other hand, the absence of autocorrelation in the data used is reflected in relatively high Durbin-Watson figures. The Modern Production Period The acreage supply response model was fitted to time series data covering the period (1960-1974). The data used included the same items mentioned in the acreage response model fitted to the traditional production period. The number of observations is 13. Five regression Table 2. The estimated acreage supply response in the traditional production period. Variables Regression 1 Parameter S. E. t -465.19 280.290 -1. 6596 +756.37 330.556 2. 2818* -322.11 +517.30 306.772 39 3.466 -1. 049 1. 315 . . + 28.415 25.952 -349.06 +541.64 331.477 419.583 x(2) = lagged relative yields + 26.360 27.906 x(3) = total annual rainfall + Constant x(l) = lagged relative prices -318. 17 +492. 10 374. 452 489.800 -0. 849 1.,004 x(2) = lagged relative yields + 24.66 30.491 0.,809 Constant x(l) = lagged relative prices /^BA I PD/t-l 2 3 4 Coefficient Constant x(l) = lage;ed relative prices / PB ] I PD/t-1 x(2) = lagged relative yields f^3-) \YD/t-l Constant x(l) = lagged relative prices Durbin-Watson R 2 i 0.0548 0.1532 1. 095 2.268 0.343 2.176 0.421 2.164 0.429 -1. 053 1. 2909 . 0. 944 0. 357 tv o Table 2. Continued. Variables Regression 4 cont. 1. Parameter Coefficient S. E. t x(3)=total annual rainfall + 0.0509 0.1638 0.3107 x(4) = lagged acreage (one period) + 0.0876 0.355 0.246 Constant x(l) = lagged relative prices -306.62 +488.21 412. 118 528.91 -0.744 0.923 x(2) = lagged relative yields +23.746 33.527 0.708 x(3)=total annual rainfall + 0.0505 0.1767 0.286 x(4) = lagged acreage (one period) + 0.1238 0.4623 0.267 x(5) = lagged acreage (two periods) - 0.06948 0.495 0.140 Durbin-Watson R 2 2.227 0.434 2.227 0.436 Notation: * Statistically significant at the 95% level. . . Statistically significant at the 80% level. Cs) 22 equations have been estimated, using a stepwise statistical procedure. The estimated acreage supply response functions are summarized in the table on the following page. The estimated coefficients of the relative prices of bread wheat to Durum wheat are positive, highly significant in each of the two first regression equations, and relatively significant at the 80% level in the remaining 3 regression equations fitted to the data. Their standard errors are quite small, and the R's squared associated with the lagged relative prices are statistically significant. The estimated coefficients of the relative yields of bread wheat to Durum wheat and the one period lagged acreage are positive but, in statistical terms, they are relatively insignificant. Estimated coefficients of the two periods' lagged acreage are positive and insignificant. The coefficient of the total, annual rainfall is unexpectedly negative, insignificant (i.e. , in statistical terms); and its standard error substantially large. Finally, the absence of autocorrelation in the time-series data used is reflected in relatively higher Durbin-.Watson figures. Table 3. The estimated acreage supply response in the modern production period. Variables Regression 1 Parameter Coefficient S. E. t Constant x(l) = lagged gged relative relat prices -332.62 +619.05 247.359 289.067 -1.344 2.142* Constant x(l) = lagged relative prices -416.53 +756.36 285.22 364.27 -1.460 2.076* Durbin-Watson . R 2 3.043 0.314 2.792 0.345 2.699 0.374 VPD/t-l x(2) = lagged acreage (one period) 0.177 0.2712 Constant x(l) = lagged relative prices -377.36 +632.53 x(2) = lagged acreage (one period) + 0.1892 0.2819 0.671 x(3) = lagged acreage (two periods) + 0.1618 0.2660 0.608 Constant x(l) = lagged relative prices -418.22 +707.90 x(2) = lagged acreage (one period) + 0.224 323.130 429.090 0.654 382.09 479.98 0.306 -1.044 1.474.. -1.094 1.474.. 0.732 tv> Table 3. Continued. Variables Regression 4 cont. 5 Parameter Coefficient S. E. x(3) = lagged acreage (two periods) + 0. 168 0.280 0.600 x(4) = lagged relative yields + 16.815 36.159 0. 465 Constant x(l) = lagged relative prices -520.81 +845.79 x(2) = lagged acreage (one period) + 0. 255 0.3262 0.783 x(3) = lagged acreage (two periods) + 0. 184 0.296 0.623 x(4) = lagged relative yields + 28.38 x(5)=total annual rainfall 1. 0. 113 438.15 556.77 42.84 0. 194 Durbin-Watson R 2.691 0. 393 2.869 0.426 -1.188 1.519 0.662 -0.584 Notation: * Statistically significant at the 95% level. . . Statistically significant at the 80% level. to 25 Yield Response The Traditional Production Period The bread wheat yields response nnodel was fitted to time series data covering the period (1949-1963). 1. The data used are: The amounts of fertilizer, conaputed as the ratio of the total quantity of nitrogen units used by the Tunisian farmers each year for bread wheat production to the total acreage harvested at the end of the year. They are expressed in units of nitrogen per hectare; 2. The acreage of bread wheat as reported by the appropriate Services in the Department of Agriculture, and expressed in 1000 hectares. 3. The national total annual rainfall as reported by the world weather report at the end of each year. The number of observations is 15. Three regression equations have been estimated using a stepwise statistical procedure. The es- timated yields response equations are summarized in the table on the following page. The estimated coefficients of the total annual rainfall are positive, relatively significant (at the 80% level), but their standard errors are quite large, and the associated R's squared low. Table 4. The estimated yield response in the traditional production period. Variables Regression Parameter 1 Constant x(l)=amounts of nitrogen per hectare 8.1327 -0.26568 0.7950977 0.2375050 10.22850 - 1.11862.. Constant x(l)=amounts of nitrogen 6.1784 -0.31684 2.1101919 0.242959 2.927881 - 1.304097.. x(2)=total annual rainfall +0.010660 0.0106623 0.999798 Constant x(l)=amounts of nitrogen 9.1185 -0.48295 6.35188979 0.4202559 1.435554.. - 1.1491758.. x(2)=total annual rainfall +0.012812 0.01184902 1.081230 x(3)=acreage of bread wheat -0.017214 0.03493137 -0.4928038 2 3 1. Coefficient S. E. Notation: . . Relatively significant at the 80% level. t Durbin-Watson R 2 0.087 1.385 0.157 1.480 0.176 27 On the other hand, the estimated coefficients of fertilizer are relatively significant (at the 80% level), but negative and their standard errors relatively large,, the associated R's squared quite low. The estimated coefficient of the acreage is negative, highly insignificant . and with.a large standard error. The explanatory power of the acreage as an independent variable is poor. Finally, the absence of a high autocorrelation is reflected in a high Durbin-Watson statistic. The Modern Production Period The yield response model was fitted to time series data covering the modern production period (1960-1974). The data used included the same items mentioned in the yields response model fitted to the traditional production period. The number of observations is 15. Three regression equations have been estimated on a stepwise basis. The estimated yields response equations are summarized in the table on the following page. The estimated coefficients of fertilizer and total annual rainfall are positive and highly significant in the 3 regression equations. Their standard errors are small and, as independent variables, the total annual rainfall and fertilizer have a highly significant explanatory power in the model since more than 50% of the yield variation is explained by fertilizer applications only, and 70% by the two factors in common. Table 5. The estimated yields response in the modern production period. Variables Regression 1 1. Parameter Coefficient S. E. Dur bin-Wat son C onstant x(l)=amounts of nitrogen per hectare 3.4796 +0.28640 0.940868 0.078726 3. 69833302 3. 6378775*= Constant x(l)=amounts of nitrogen 0.38950 +0.20810 1.41808757 0.07196562 0.27466605 2.8916735** x(2)=total annual rainfall +0.016182 0.00619610 2.612038** Constant x(l)=amounts of nitrogen -0.98619 +0.19963 1.9783517 0.07247994 -0. 4984909 2.7542746** x(2)=total annual rainfall +0.014922 0.006323865 2.359590** x(3)=acreage of bread wheat +0.0095770 0.0096011 0.9974898. . 1.486 0.504 1.387 0.684 1. 339 0. 710 Notation: ** Statistically, highly significant. . . Statistically significant at the 80% level. co 29 On the other hand, the estimated coefficient of the acreage is positive, but relatively insignificant. Its standard error is quite large and the R squared associated with the acreage, as an independent variable in the model, is poor. The Demand for Fertilizer The Traditional Production Period The demand for fertilizer by Tunisian farmers was fitted to time-series data covering the period (1953-1963). 1. The data used are: The wholesale prices of bread wheat and fertilizer as reported by the appropriate service in the Department of Agriculture, expressed in millimes per quintal; 2. The farm income measured by the per capita income of the Tunisians as reported by The United Nations Statistical Yearbooks. It is expressed in dinars (1000 millimes) per year. The number of observations is only 10, due to lacking data for the period (1949-1952). Two regression equations have been estimated on a stepwise basis. The regression equations are summarized in the table on the following page. The estimated coefficients of the relative price of bread wheat to fertilizer are positive and significant in both regression equations at Table 6. The estimated demand for fertilizer in the traditional production period. Variables Regression 1 Parameter S. E. t -15.863 9.600150 Relative prices +16.053 7.9913200 Constant -27.118 Relative prices +16.132 8.6043411 1.8748852* (PF)t Lagged Income +0.062044 0.14706532 0.42188019 rZBjt-1 28.609728 Durbin-Watson -1.652381 Constant fPB)t-l 2 Coefficient 2.0087801* R 2 0.4021 1.720 -0.9478620 1.828 0.422 M t-1 1. Notation: * Statistically significant at the 95% level. o 31 the 95% and 90% level respectively. Their standard errors relatively small and the explanatory power of the relative prices, as an independent variable, is statistically significant. However, the estimated coefficient of the lagged income is insignificant but positive; Its standard error relatively large and the R squared associated with the lagged income low. The absence of autocorrelation is reflected in a high Durbin-Watson statistic. The Modern Production Period The demand for fertilizer in the modern period was fitted to time-series data covering the period (1960-1974). The data used included the same items to which the demand for fertilizer in the traditional period was fitted. The number of observations is 15. Two re- gression equations have been estimated on a stepwise statistical procedural basis. These regression equations are summarized in the table on the following page. Table 7. The estimated demand for fertilizer in the modern production period. Variables Regression 1 2 Parameter Coefficient S. E. 10.5155625 t Constant -27.388 Lagged Income +0.17314 Constant -35.228 Relative Prices (PB)t-l +4.0218 6.73741149 0.5969300 lagged Income +0.18728 0.05483280 3.40929397** Durbin-Watson 2 -2.604477 0.0483187465 3. 58328551**1 16.9959678 R 1.84 0.4969 2.747 0.5114 -2.072736 fPF)t 1. Notation: ** Statistically, highly significant. 33 IV. INTERPRETATION OF THE ESTIMATED RESULTS The Acreage Supply Response The Traditional Production Period The initial hypothesis that Tunisian farmers respond to the relative prices of bread wheat to Durum wheat, in their acreage allocation decisions, is supported by the empirical results. Q However, it is worthwhile to notice that, during this period, the climatic conditions and essentially the rainfall were to a certain extent of major importance in the eyes of the Tunisian farmers in their acreage allocation decisions, as stated by MM. John D. Hyslop and Reynold P. Dahl (10.). In fact, most of the producers, if not all of them, are usually influenced within certain limits, by the first rainy days of the seeding season in November following the dry period of summer. But, due to high uncertainty about the weather forecasts in the area and the very marked inter- and intra-seasonal variability of the rainfall, the response of farmers to the rainfall was not reflected in the regression analysis in which crop-year rainfall was approximated by total annual rainfall, even though the estimated coefficients are still positive. g Statistically speaking, the lagged relative prices are the only significant variable in the estimated regression equation. 34 On the other hand, the regression analysis reflected satisfactorily the fact that, after the Second World War, Tunisian farmers felt the,need for more wheat production and there was a transition from the traditional rotation idle land - wheat to the actual rotation of bread wheat on bread wheat, as was stated by the two previously mentioned researchers. However, it might partly be due to the relative high.collinearity between the one and two periods lagged acreages (r = O.SSl), that the response of the farmers to them was relatively insignificant. In fact, . eliminating the two periods lagged acreage from the model and increasing the number of observations from 1 3 to 14*.-;as illustrated in the table below, the estimated coefficients of the ;One period lagged acreage were positive and essentially more indicative. ..Also,' it is'worthwhile, to.notice that, by dropping the two periods lagged acreage, the farmers' response to the lagged relative price of their output and the total annual rainfall was satisfactorily improved in the model fitted to this'period. In addition, the response of Tunisian farmers to the lagged relative yields, as reflected in the regression analysis, was relatively insignificant partly due toithe considerations of the climatic conditions by; most pf the farmers." during this period. Table 8. The revised estimated acreage supply response in the traditional production period. Variables Regression 1 2 Parameter Constant x(l)=lagged relative prices \PD/t-l Constant x(l) = lagged relative prices Coefficient S. E. t -425.96 +712.66 242.0719 286.4632 -1.7596 2.4877** -473.59 +748.60 255.5030 296.2350 -1.8535 2.5270** Durbin-Watson R 2 1.762 0.322 1.779 0.350 1.980 0.383 (•F§)t-1 3 x(2)=total annual rainfall + 0.086634 0.12078 Constant x(l) = lagged relatiy.e .prict tive .prices -390.50 +602.67 x(2)=total annual rainfall + 0.09599 0.12348 0.77734 x(3) = lagged acreage of bread wheat + 0.21850 0.283838 0.769807 Constant x(l) = lagged relative prices -294.75 +457.60 281.4685 356.0536 -1.3873 1.69262. (M VPD/t-l 348.4856 468.6750 -0.84581 0.97637 Table 8. Continued. Variables Regression 4 cont. 1. Parameter Coefficient S. E. t x(2)=total annual rainfall + 0.08875 0.128729 0.689444 x(3) = lagged acreage of bread wheat + 0.23328 0.29548 0.78946 x(4) = lagged relatlve yields + 12.661 25.2390 Notation: ** Statistically, highly significant. Significant at the 90% level. 0.501628 Durbin-Watson 1.980 R 2 0.398 37 The Modern Production Period The initial hypothesis that the Tunisian farmers are responsive to the relative prices of bread wheat to Durum wheat is highly supported by the empirical, evidence during this period too. But it should be made clear that the absence of a significant difference between the Tunisian farmers' response to the lagged relative prices in the traditional and the modern periods of production might partly be explained by the relative high correlation between the two periods lagged acreage and the lagged relative prices (r = 0.600). 9 Eliminating the two periods lagged acreage from the model and increasing the number of observations from 13 to 14 was undertaken in this study. As a re- sult, the estirnated coefficients of the lagged relative prices of bread wheat to Durum, wheat appeared to be highly significant in the four regression equations, their standard errors relatively small, and the R's squared associated with the first two regression equations were statistically significant, as illustrated in the following table. In addition, as was mentioned earlier, due to the high instability of the climatic conditions and the marked interT and intra-seasonal •9 In Elements of Econometrics, Jan Kmenta mentioned that, when there are more than 2 independent variables in the model, the low magnitude of the correlation coefficients (r=0. 50 in :his example) are sometimes underestimating the high collinearity between the pairs of independent variables. For further information, see Jan Kmenta, (13, p.. 482)'.' Table 9. The revised estimated acreage supply response in the modern production period. Variables Regression 1 2 3 4 Parameter Constant x(l) = lagged relative prices Coefficient S. E. t -475.08 +772.96 256.47549 300.54025 -1.8532 2.571899** \PD/ t-1 Constant x(l)=lagged relative prices -593.09 +863.39 289.1093 318.5732 -2.0514 2.71019** VPD/ t-1 x(2) = lagged relative yields +30.300 lYD/t-l Constant -669.48 x(l)=lagged rela- +1001.8 tive prices x(2)=lagged relative yields +39.873 x(3)=total annual rainfall - 0.1024 Constant -704.01 x(l)=lagged rela- _1009. 20 tive prices 33.3784 354.3211 412.3305 38.47958 0.1847327 394.49719 483.4049 0.90776 Durbin-Watson _ R 2.178 0.337 2.106 0.379 2.085 0.396 -1.974138 2.42949** 1.036212 -0.554291 -1.784583 2.08759* 00 oo Table 9. Continued. Variables Regression 4 cont. Parameter Coefficient S. E. 40.36315 Durbin-Watson x(2) = lagged relative yields +39.900 x(3)=total annual rainfall - 0.10218 0.1938430 -0.527113 x(4) = lagged acreage of bread wheat + 0.010169 0.2970881 +0.0342285 1. Notation: ** Highly significant. . . Significant at the 95% level. 2. Statistically speaking, regression (3) is significant. R 0.988529 2.078 0. 396 00 40 variability of the rainfall in the country, the response of the Tunisian farmers to the rainfall was insignificant and even negative in the regression analysis in which crop year rainfall was approximated by total annual rainfall. On the other hand, with the wide extension programs undertaken by the Tunisian Department of Agriculture urging the farmers to grow more new high yielding varieties of bread wheat, the positive response to the lagged relative yields seems to be encouraging even though not highly significant during the beginning of this period. The Yield Response The Traditional Production Period The negative effects of nitrogen application and the acreage on the yields, as reflected in the regression analysis, seem to be puzzling, but historical evidence shows that they are not. In fact, during this period, the average acreage devoted by the Tunisian farmers to Durum wheat relative to bread wheat was more than 91%, as is illustrated in the table on the following page, and bread wheat was grown on very marginal lands except on European farms. In addition, it is worthwhile to notice here the poor contribution of fertilizer to the explanation of the yields variability. On the other hand, even though the effect of rainfall was biased for the same reason Table 10. The acreage of wheat on Tunisian and European farms during the period (1949 - 1958). Tunisian Farmers Harvest Year Area in Area in Durum Wheat Bread Wheat (1000 hectares) European Farmers Percent of Total in Durum Wheat Area in Area in Durum Wheat Bread Wheat (1000 hectares) Percent of Total in Durum Wheat 1949 604 55 91.7 64 108 37.2 1950 443 53 89.3 85 115 42.5 1951 735 67 91.6 100 107 48.3 1952 830 78 91.4 122 126 50.8 1953 757 64 92.2 116 120 49.2 1954 1021 74 93.2 132 131 50.2 1955 690 65 91.4 144 123 53.9 1956 820 90 90.1 145 133 52.2 1957 958 83 92.0 135 119 53.1 1958 976 74 93.0 133 99 57.3 Source: Republic of Tunisia, Secretarial of State for plan and for the national economic and Annuaire Statistique de la Tunisie, various issues. t^. 42 mentioned earlier in this study, the positive sign of the estimated coefficients reflects its impact on the yields, specifically during this period. The Modern Production Period The response of the new high potential varieties to nitrogen applications and rainfall is reflected by the time-series data fitted to this period. In fact, more than 50% of the yields variability are ex- plained by nitrogen application and about 70% by the combination nitrogen - rainfall. In addition, it is of interest to notice here, that the positive sign of the estimated acreage coefficient reflects the recent tendency of the Tunisian farmers to shift more of their potential lands to the new varieties of bread wheat introduced in the country under3 the urgence of the Tunisian Government. The Demand for Fertilizer The Traditional Production Period The positive estimated coefficients of the relative prices of bread wheat to fertilizer, in both regression equations, reflect a very interesting result, in the sense that the Tunisian farmers were responsive in their purchases of fertilizer, to these relative prices. 43 In addition, the regression equations fitted to this period showed the positive effect of income on the Tunisian farmers' purchases of fertilizer, even though the estimated coefficient was relatively insignificant. The Modern Production Period The hypothesis that there is a positive effect of farm-income and price on the purchases of fertilizer by Tunisian farmers is supported by the empirical results related to this period, but, the magnitude and the significance of the estimated coefficients related to the two periods of production, in a certain way, seem to be misleading. This might partly be due to the difference in the number of observations fitted to each of these two periods. In fact, the number of observations used in the traditional period was only 10, compared to 15 observations in the modern production period. The reliability of the modern period estimates seems to be plausible. Furthermore, it is worthwhile to notice that, statistically speaking, the income-relative price effect is highly significant at the 95% level, and with 12 degrees of freedom. In fact, in the traditional period, the estimated coefficients of the relative prices were statistically significant while in the modern period they were not. 44 Comparison A comparison of the empirical results underlying each of the tw<3 major periods of production in the country reveals that: 1. In the recent yearst the Tunisian farmers seem to be relatively more responsive, *■ in their acreage allocation decisions, to the relative prices (i. e. of bread wheat to Durum wheat) than they were in the traditional period; 2. The response of yields to fertilizer applications in the modern production period is substantially higher^ than their response in the traditional period; 3. And, that in both periods, the demand for fertilizer was highly related to farm income and the relative prices of bread wheat to fertilizer. l-'-The estimated coefficients of the relative prices related to the two periods are not statistically different at the 95% level, but significant at the 80% level after the exclusion, of the two- periods lagged acreage from the model. l^The estimated coefficients of fertilizer, in the two periods, are significantly different. 45 V. POLICY IMPLICATIONS The production of bread wheat in Tunisia has been a highly traditional agricultural activity for decades, although it was not as important as was the production of Durum wheat. 13 However, in the last decade, with a flourishing domestic market and a high importation trend, this agricultural commodity has attracted the attention of many governmental agencies, researchers and producers as well. As a matter of fact, Tunisia was one of the first countries to try to change the structure of the production pattern of this agricultural commodity in the country by a progressive transition from a traditional agricultural sector, reliant to a high degree on economically uncontrolled factors of production, to a modern sector under which some of the major factors of production are controlled (16). However, some of the practical limitations to the increase of production in the country need more attention, essentially concerning the Tunisian farmers' response to the prices of output in their acreage allocation decisions, and the response of the new high yielding varieties to fertilizer applications relative to the traditional local varieties under similar climatic conditions. l-^In fact, the average relative acreage of Durum wheat to bread wheat was more than 91% for the period (1949-1958). 46 The empirical analyses covered by this study supports the hypotheses: 1. that the Tunisian farmers are responsive to the relative prices of bread wheat to Durum wheat, (among others), in their acreage allocation decisions; 2. that the new high yielding varieties, which were introduced in the modern production period, are more responsive to fertilizer applications, (among others), than the local varieties and that the demand for fertilizer by the Tunisian farmers is highly responsive to both farm-income and the relative prices of bread wheat to fertilizer. This implies a possibility that an appropriate price policy in favor of bread wheat would contribute relatively more to achieving increased domestic production with respect to an expected increase in both the acreage and the yields. The Acreage of Bread Wheat The stagnation of bread wheat acreage at about 1/5 of the acreage of Durum wheat for more than 25 years, as is illustrated in the table and figure on the next two pages, could be true in the times when Tunisia was among the major exporting countries of Durum wheat in the world grain market, and the internal prices of bread wheat were relatively low. (5). 1400 1300 1200 Durum Wheat 1S1100 H rtlOOO -\ -M « 900 o 800 o 2 700 .2 600 -\ M 500 H Z 400.. H 300 5 Bread Wheat 200 100 -I T !—l 1 1 1—i 1 1 1 1 1 1 r 1949 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 Figure 1. Year Acreage of bread wheat and Durum wheat during the period (1949- 1974). -j 48 Table 11. The average acreage of wheat in Tuiiisia (1949-1974). Average Acreage 1949-1969 Average Acreage 1959-1969 Average Acreage 1960-1974 (1000 ha) Durum Wheat 874 8 48 855 Bread Wheat 170 154 176 1044 1002 1031 Total Ratio: Durum Wheat Bread Wheat The Average Ratio: Source: 5. 14:1 5.51:1 4.86:1 5. 17:1 Production de ble en Tunis ie, tendences et variabilites. However, with the'recent: 1. vanishing privileges of Tunisian Durum wheat in the French market as well as in the world market; 2. higher internal prices of bread wheat; 3. and, the higher population growth the Tunisian exports of Durum wheat disappeared, as-it is illustrated below, and the need for a revision of the-internal pricing piOlicy in favor of bread wheat seems to: be in. actual evidence. Even though the domestic demand for Durum wheat is highly inelastic, 14 the possibility of directly stimulating Tunisian farmers to allocate more of their lands to bread wheat production by the means M. John D. Hyslop found an average pif ice elasticity of (-0. 2) in a projection analysis of wheat consumption in Tunisia. 49 Table 12. The Tunisian exports of Durum wheat 1957-1966. (1000 tons) 1957/58 1958/59 1959/60 1960/61 1961/62 1962/63 1963/64 1964/65 1965/66 1966/67 1967/68 1968/69 , 129 154 150 87 17 47 132 17 103 1 1 0 of higher relative prices (i. e. of bread wheat to Durum wheat), other things being equal, could be highly practical, in recent years, for two main reasons: 1. The recent introduction of high yielding varieties of Durum wheat in the country; 2. And the expected higher income levels of Tunisian consumers. In fact, during the modern production period, high yielding Canadian and American varieties of Durum wheat were successfully developed, 1 5 and as it was reported in a recent statistical survey (25), the yields of the new varieties are significantly superior to the local varieties under similar growth conditions. 1 See also my own empirical results developed in Appendix B of this study. 50 Consequently, with an increasing care of the Tunisian farmers about: 1. the.application of sufficient amounts of fertilizer in Durum wheat production; 2. the w^eed control which 'is a major problem in the intensive areas of production; 3. and, the application of a proper crop rotation the switch of an expected higher acreage from Durum wheat to bread wheat would probably be compensated by expected higher yields of Durum wheat. On the other hand, -the-expected price increase would probably be reflected in a higher farm-income le.vel which inneans a higher consumer income (since more than 60% of the Tunisian population are engaged in agriculture), and more bread wheat production; since in a recent statistical survey, M. John D. Hyslop found that the consumption of bread wheat versus Durum wheat, was highly related to the income level of the consumers. In fact, as it is illustrated in the table below, the consumption of bread wheat in the large cities (higher income levels) was reported to be significantly higher than in the small counties (lower inco>me levels). 51 Table 13. The estimated per capita consumption of wheat by degree of urbanization in Tunisia (1966) (in kilograms). Cereals Large cities Counties Others Total Durum wheat 50.0 88.1 99.0 85.2 Bread wheat 80.0 70.1 60.0 66.7 Source: Une analyse de quelques politiques possibles de ble en Tunisie. Report of a research in Agricultural Economics #3, by John D. Hyslop, University of Minnesota. In addition, a recent analysis of the demand for wheat in Tunisia undertaken by the same economist cited above aiujl who was working on the projection of wheat production in Tunisia (12), showed that, beyond a certain level of income (estimated at about 200 U. S. dollars per capita); 1. the per capita consumption of durum wheat was inversely related to the individual income level; 2. and the per capita consumption of-btread wheat positively related to the income level, as it is■ illustrated in the projection table on the next page. This implies that any relative "price increase would certainly require more attention to a further increase of bread wheat acreage in the country. The Yields of Bread Wheat The possibility of a revision of the pricing policy in favor of 52 Table 14. Projection of the production of cereals in Tunisia. Per Capita Consumption of Wheat (kgs.) Durum wheat (kgs • ) 1966 66.7 85. 1 1968 69.9 84. 7 1970 73. 1 84. 0 1972 76.1 83. 2 1974 79. 1 82. 0 1976. , 81. 7 80. 7 1978 84. 5 79. 3 1980 8-7.0 77. 5 1982 89.3 75. 6 : 1984 91. 4 73. 5 1986 93. 4 71. 2 Period Bread wheat Source: Une analyse de quelques politiques de b'le possibles en Tunisie. Report of a research in Agricultural Economics #3, by John D: Hyslop, University of Minnesota. bread wheat, other things being equal) on the other hand, is expected to have a positive impact on the domestic yields with respect to: 1. The expected stimulus to fertilizer applltatibn by the Tunisian farmers; 2. And, the increase of their individual incorrie levels. In fact, on the basis of the empirical results "covered by this study, the demand for fertilizer by the Tun'isiaii farme'rs was revealed to be positively responsive to the relative prices of bread wheat (to fertilizer), and the new varieties, recently introduced in the country, 53 highly responsive to fertilizer applications. This implies that an appropriate pricing stimulus would be reflected in a substantial improvement of yields. This indi'rect positive impact on yields is highly suggestive and should not be overlooked insofar if the main objective is to achieve an increasing domestic production. On the basis of the empirical results discussed in Chapter II of this study, the demand for fertilizer by Tunisian farmers was revealed to be highly determined by farm-income. In fact, about 50% of the variation in the quantities demanded by farmers were explained by. variations in their income. This implies that, in-a developing nation like Tunisia with a relatively high rural population (60% of the total population), any expected revision of the internal pricing, policy in favor of this basic agricultural commodity (i. e. bread wheat) would eviden.tly improve farm-income, thereby increasing the use of fertilizer, and consequently, other things being equal, the dbmestic yields would substantially improv;e due to the high response of the new varieties successfully developed in the area. It should be made clear at this stage, that aTnScessary time of adjustment to any appropriate price revision should1 be expected, and achieving an equilibrium domestic production level is a long-run •adjustment procedure. On this matter, a leading' American agricul- tural economist, John D. Black can be quoted (3): 54 . . . the full effects of a definite change in relative prices for a given product will take several years to appear. The fact of the time of the response is fully as important as the amount of it. If the change in prices is merely temporary, lasting for two years, let us say, the response will be headed off before it has happened. In fact, the empirical results developed in this study can be helpful in determining the length of time required to make full adjustment to a change in relative wheat prices. During the modern production period, other things being the same, the actual acreage devoted each year by Tunisian farmers to bread wheat production is determined by the lagged relative price of bread wheat to that of Durum wheat and the one period lagged acreage such that: Numerically expressed as: (1) where the coefficient of price expectation, p, is equal to 0.8Z3 (P = 1 - 0. 177). Assuming that, initially (t = 0), Tunisian farmers /PB\ have reached a long-run equilibrium at a relative price j ——I of 0.82 \ D/ which has been held constant for a sufficiently long time, we could write: This is the simplified form of the original general Nerlovian Model after the elimination of the two-periods lagged acreage (i. e. , the coefficient of adjustment, -y, is equal to one), because this latter did not make a significant contribution in the model. 55 the acreage equilibrium would then be such that: (2) or: x = 247. 49 . Suppose now that after a certain price revision in favor of bread wheat a new relative price of 0.89 was set once and for all so as to stimulate domestic bread wheat production; in period t = 1, the Nerlovian expected price of bread wheat to that of Durum wheat would be such that: B 1* B * D/l D/ o (P + P B \p«) D/ o \\pJ D/ o or: B I* 0.82 + 0.823 (0.89 - 0.82) = 0.878 D /I Replacing f—— I and x D x by their respective values in (1), we have: = 416.5 3 + 0. 177 (247. 49) + 756. 36 (0.8 78) or: x1 = 291.065 Similarily, in period t = 2, the expected relative price of bread wheat to that of Durum wheat would be: _B D/l or: J3 I* + P D/ 1 D/I X Path of adjustment before the price decrease CO (0 U nl 310 -tj o <u X! o o o n) a) 300 - 290 - Path of adjustment after the price decrease 280 270 nl ID !H 260 - 250 - ^3 0) bJO ni 0) 240 o < ~ 2 30 r 220 210 > 1_ 0 Figure 2. T- 3 4 7 Bread wheat acreage adjustment to relative price changes. Ul 5? ' =0.878 +0.823 (0.89 - 0.8 78) =0.888 Replacing! ——I and x x in (1), we get: = -416. 63 + 0. 177 {291. 065) + 756. 36 (0. 888) or: x = 306. 49 This average adjustment estimation has been carried out for five periods as illustrated in the table below and Figurte 2. From (2), the hew equilibrium would be: xV = -416. 53 + 0. 177 x* + 756. 36 (0. 89) or: x* = 311.82 Table 15. Acreage adjustment to changes in relative prices (a). Period (t) Relative[_B] Price l^n/t ^ ' " Expected / B\ * Relative I'P*'^. I. Price ^ ' Acreage (x ) % of adjustment 0 1 2 3 4 0.82 0.89 0.89 0.89 0.89 0.8200 0.8780 0.8880 0.8896 ' 0.8899 247.490 291.065 306.490 310.570 311.520 67.74 91.71 98.06 99.53 n 0.89 0.8900 311.820 100.00 58 This indicates that an increase of 7% in the relative price of bread wheat could lead to a 26% increase in acreage in the long run. It is significant, however, to note that 67. 74% of the adjustment to the new long run equilibrium would be accomplished in the first year following the change in relative price and 99. 53% within the first four years. However, the effects of this price increase could be aborted before its full effect has been attained. In fact, suppose that in period t = 2, /PB\ the relative price \——I temporarily falls to 0. 85, due to, say, a high \ D/ domestic supply of bread wheat. An expected lower equilibrium level will be sought by farmers and the path of adjustment will tend to deviate from its initial course depending upon the magnitude of price change as is illustrated in the table below. Table 16. Period (t) Acreage adjustment to changes in relative prices (b). Relative/ Bj PriCe P i D/t-l Expected/ Relative I Price ^ B )* Mt Acreage (xt) 0 1 2 3 0.820 0.890 0.850 0.890 0.820 0.878 0.855 0.884 247. 49 291.06 281. 67 301. 94 n 0.890 0.890 311.82 However, there is a possiblity .of accelerating the attainment of such an equilibrium, and this is by improving the training programs of the Tunisian agricultural extension agents and providing them with a 59 relatively large scope of action throughout the country so as to create an effective propaganda which was described by John D. Black as an important factor for the success of any governmental production program when he wrote-(3): ... it is surely true that as time goes on, and this (propaganda) and related problems are better understood, and farmers learn from experience that production programs recommended to them by responsible extension agencies are well founded, acreages will be greatly influenced by such means. Finally, it should be stressed that this study was not supposed to cover the whole problem and additional work on this subject is recommended by the author. 60 BIBLIOGRAPHY 1. Bean, Louis H. , "The Farmers Response to Price. " Journal of Farm Economics, vol. II (1929), pp. 368-85. 2. Behrman, J. R. , Supply Response in Underdeveloped Agriculture, Amsterdam, 1968, p. 1-18 and p. 151-170. 3. Black, J. D. , Elasticity of Supply of Farm Products, Journal of Farm Economics, April, 1924, vol. VI, p. 145-155. 4. Closer Farm-Factory Tie Urged on Poorer Lands by World Bank, New York Times (New York), p. 5, February 26, 1966. 5. Dahl, R. P. , International Trade and Price Prospects for Cereals and Their Implications to Tunisia, Marketing Report no. 5, January, 1970. 6. Ferguson, C. E. , Microecononaic Theory, Third Edition, Richard D. Irwin, Inc. , 1972. 7. Hubbard, W. B. , Cotton and the Cotton Market, D. Appleton & Co. , 1923, p. 20. 8. Hyslop, J. D. , Analyse de politiques possibles de production Cerealiere en Tunisie, Rapport de Recherche en Economie Agricole no 3, Juin, 1971. 9. Hyslop, J. D. and Dahl, R. P. , Prix du ble et Politique de Prix en Tunisie, Rapport de Recherche en economie agricole no 2, Avril, 1970. 10. Hyslop, J. D. and Dahl, R. P. , Production de ble en Tunisie: Tendences et variabilities, Rapport de recherche en economie agricole no 1, Mai; 1970. 11. Hyslop, J. D. and Dahl, R. P. , Wheat Prices and Price Policy in Tunisia, Staff paper, University of Minnesota, June, 1970. 12. Hyslop, John D. , The Tunisian Cereals Sector: An Exannination of Production, Prices, and Some Alternatives for the Future, Institute of Agriculture, University of Minnesota, June, 1970. ■'61 13. Kmenta, Jan, Elements of Econometrics - New York, 1971. 14. Nerlove, Marc, The Dynamics of Supply: Estimation of Farmers' Response to Price, Baltimore, 1958. 15. Owen Tvirak, Factors Determining the Price of White Wheat in the Pacific Northwest, Washington State' lifrilversity, November 18, 1974. 16. Purvis, M. J. , "The New Varieties Under Dry-land Conditions: Mexican Wheats in Tunisia," Ameriiian Journal of Agricultural Economics, vol. 55, no 1, February, 1973. 17. Republic of Tunisia, le journal officiel de la Republique Tunisienne-Lois et Reglements - Prix des cereales - June, 1974. 18. Samuelsbn, Paul A. , Economics, An'Iritrbductory Analysis, Sixth edition. New York, 1964. 19. Schultz, T. W. , Economic Crises in World Agriculture, the University of Michigan p^'ess. September, 1964. 20. Schultz, T. W. , Food for'the World. Press, 1945. 21. Schultz, T. W. , Production and Welfare of Agriculture, New York, the MacMillan Company, 1949. 22. Smith, B. B. , "Forecasting the Acreage of Cotton, " Journal of American Statistical Association, 1925, vol. 20, p. 31-47. 23. Snedecor, George W. and Cochran, William G. , Statistical Methods - Sixth Edition, Iowa State University, 1937. 24. Tunisian Ministry of Agriculture - Budget Economique 1973, la production. 25. - Bureau du Plan et du developpement agricole (B. P. D. A. ), enquetes statistiques, 1974. 26. _^ - Comptes ressources-emplois des produits de 1'agriculture et de la Peche (1964-1968). University of Chicago 62 27. Tunisian Ministry of Agriculture (1970-1972). 28. - Prix moyens des produits de 1'agriculture et de la Peche (1965-1972). APPENDICES 63 APPENDIX A: BASIC DATA 64 Appendix Table 1. Period 1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 I960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 The acreage and yields of bread wheat and Durum wheat in Tunisia (1949-1974). Acreage of Bread Wheat (1000 hectares) Acreage of Durum Wheat (1000 hectares) 170 170 170 200 180 200 180 220 210 180 180 190 110 100 170 180 180 160 180 180 145 280 200 260 230 195 675 520 830 975 875 1175 825 1100 1100 1150 1150 800 700 930 900 890 675 675 600 700 600 750 700 1000 1020 1046 / Bread Wheat (qx/ha) 11.0 10.8 7. 1 10.8 11.0 9.3 5.5 6.3 6.0 5,2 5.0 4.0 3.8 6.8 8.0 4.7 5.5 3.0 2.5 5.5 5.5 5.4 10.0 9.9 10.2 10.4 Durum Wheat (qx/ha) 5.5 5.4 2.5 5.0 4.5 3.8 3.5 3.4 3.3 3.7 3.6 3.3 2.2 4.0 5.3 3.5 4.0 3.8 3.8 4.0 3.8 4.9 5. 7 7.5 6.7 6.6 / Source: Production de BLE EN TUNISIE, Tendences et Variabilites 65 Appendix Table 2. Period 1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 Source: The yields of bread wheat and the total annual rainfall in Tunisia (1949-1974). Total Annual Rainfall (Millimeters/year) 150 150 155 303 233 136 66 139 232 250 256 191 170 195 265 250 320 94 150 270 300 172 200 314 350 350 Yields of Bread Wheat (quintaux/hectare) 11.0 10.8 7. 1 10.8 11.0 9.3 5.5 6.3 6.0 5.2 5.0 4.0 3.8 6.8 8.0 4. 7 5.5 3.0 2.5 5.5 5.5 5. 4 10.0 9.9 10.2 10. 4 World Weather Records and Monthly Climate data for the world. Appendix Table 3. Period 1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 I960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 The wholesale prices of bread wheat. Durum wheat and nitrogen in Tunisia (1949-1974). Wholesale Prices of Bread Wheat (Mil limes /quintal) 2500 2600 3600 3600 3600 3400 3400 3450 3450 3596 3450 3450 3450 3450 3450 3450 3450 3450 4300 4300 4300 4300 4300 Wholesale Prices of Durum Wheat (Millimes/quintai) 2932 3172 4140 4140 4140 3910 3910 3967 3967 4468 4200 4200 4200 4200 4200 4200 4200 4200 4800 48 00 48 00 4800 4800 Wholesale Prices of Nitrogen (Millimes/quintai) 3140 3160 3160 3240 2493 2677 2677 2677 28 30 3616 3616 3100 4600 3000 3000 3000 Appendix Table 3. Period 1972 1973 1974 Continued. Wholesale Prices of Bread Wheat (Millimes /quintal) 4300 4300 5300 Wholesale Prices of Durum Wheat (Millimes /quintal) 4800 4800 5800 Wholesale Prices of Nitrogen (Millimes/quintal) 3000 5000 5000 Not officially reported Source: Republic of Tunisia, Secretarial of State for plan and for the national economic and Annuaire Statistiques de la Tunisia, various issues. 68 Appendix Table 4. The yields and prices of bread wheat relative to the yield and price of Durum wheat and to the price of fertilizer (1950-1974). Period The Relative Yields lTD/t-1 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 I960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 Note: 2.00 2.00 2.85 2.27 2.58 2.58 1. 72 2.04 2.12 2. 12 1.78 1.33 1.75 1.72 1.51 1.35 1.23 0.91 0.71 0.91 1.11 1. 11 1.75 1.54 1.54 The elative Prices Dlt-l 0.85 0.82 0,87 0.87 0.87 0.87 0.87 0.87 0.87 0.80 0.82 0.82 0.82 0.82 0.82 0.82 0.82 0.82 0.89 0.89 0.89 0.89 0.89 0.89 0.89 The table is derived from previous data. The Relative Prices fPB) t-1 (PF) t - 1.08 1.09 1.09 1.11 1.38 1.29 1.29 1.29 1.22 0.95 0.95 1.11 0.93 1.43 1.43 1.43 1.43 0.86 0.86 69 Appendix Table 5. Period 1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 I960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 The estimated per capita income in Tunisia (1949-1974). Estimated Total Population (1000) 3416 3470 3500 3560 3630 3680 3745 3780 3815 3852 4105 4168 4254 4290 4304 4355 4675 4787 5067 5194 5300 5432 5568 Estimated Gross National Product (Million dinars)* 240 246 233 249 238 270 258 283 315 320 337 358 406 399 409 444 441 48 4 544 Per Capita Income (U.S. Dollars) 181 183 171 181 171 192 173 187 203 205 215 226 239 229 222 235 229 245 269 - Source: United Nations - Statistical Yearbook - different issues. *1 Dinar = Tunisian currency = 2. 75 U. S. dollars. - Not officially reported. Appendix Table 6. Annual yields and per hectare nitrogen use in bread wheat production during the period (1949-1974). Period Production of Bread Wheat (1000 T) Yields (qx/ha) 1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 I960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 180 180 120 230 200 190 100 150 140 130 120 90 43 93 130 95 120 70 75 73 80 150 11. 0 10.8 7. 1 10.8 11.0 9.3 5.5 6. 6. 7. 6. 4.0 3.8 6.8 8.0 4. 7 5. 5 3.0 2.5 5.5 5.5 5.4 Quantity of Nitrogen Used in Bread Wheat Production IT/U-F. ) 100 78 113 153 103 91 76 94 84 73 480 369 531 947 1097 1281 1385 915 1591 899 1680 2717 Acreage of Bread Wheat (1000 ha. ) 170 170 170 200 180 200 180 220 210 180 180 190 110 100 170 180 180 160 180 180 145 280 Quantity of Nitrogen (U. F. /ha) 0.588 0. 458 0.664 0.765 0.572 0.455 0.422 0.427 0.400 0.405 2.666 1.942 4.827 9.470 6.453 7. 116 7.694 5. 178 8.8 38 4.994 11.586 9. 703 o Appendix Table 6. Continued. Period Production of Bread Wheat (1000 T) 1971 1972 1973 1974 200 258 235 202 Yields (qx/ha) 10.0 9.9 10. 2 10. 4 Quantity of Nitrogen Used in Bread Wheat Production (T/U- F. ) 4400 2072 3538 5280 Acreage of Br ead Wheat (1000 h;a.) 200 260 230 195 Quantity of Nitrogen (U. F. /ha) 22.000 7.969 15.382 27.076 Source: Production de BLE EN TUNISIE, Tendences et Variabilites and FAO, production yearbook, various issues. 72 APPENDIX B: ESTIMATION OF ACREAGE AND YIELD RESPONSES FOR DURUM WHEAT EMPIRICAL RESULTS AND COMMENTS 73 Durum "Wheat Acreage Supply Response The Traditional Production Period The Nerlovian-type acreage response model discussed earlier 18 (i. e. , in Chapter II of this study) was fitted to time series data covering the period (1949-1963). 1. The data used are: The wholesale price of bread wheat relative to that of Durum wheat expressed in millimes per quintal, which are reported by the appropriate service of the Tunisian Department of Agriculture at the end of each harvesting year; 2. The total annual rainfall as the national total annual rainfall in millimeters, which is reported by the world weather report at the end of each year; 3. The acreage of Durum wheat harvested each year in 1000 hectares, and reported either by the Department of Agriculture or the Department of National Economy; 4. The national annual average yields computed as the ratio of the total annual production to the acreage harvested in the country. They are reported by the appropriate service in the Department of Agriculture in quintaux per hectare. l°The model indicated here, is the simplified form of the Nerlovian general model obtained after the elimination of the two-periods lagged acreage which was highly correlated with the one period lagged acreage. 74 The number of observations is 15. Four regression equations have been estimated using a stepwise statistical procedure. The estimated acreage supply response functions are summarized in the following table. The estimated coefficients of the lagged acreage and rainfall are positive and significant, their standard errors quite small, and the explanatory power associated with these two variables is significant. However, the estimated coefficients of the lagged price of bread wheat relative to that of Durum wheat are positive and insignificant, indicating the indifference of Tunisian farmers, in their acreage.alloca19 tion decisions, to changes in relative prices of their output. The absence of autocorrelation in the data used is reflected in a higher Durbin-Watson Statistic. The Modern Production Period The acreage supply response model was fitted to time series data covering the period (1960-1974). The data used included the same items mentioned in the acreage response model fitted to the traditional production period. The number of observations is 15. Four regression equations have been estimated using a stepwise statistical procedure. 19 In part, this result stresses the period of "production for subsistence" which arose in the country after the Second World War. Appendix Table 7. Variables Regression The estimated Durum wheat acreage supply response in the traditional production period. ^_^ Parameter Coefficient S.E. Durbin-Watson R< C onstant Lagged Acreage of Durum Wheat Constant Lagged Acreage of Durum Wheat Total Annual Rainfall C on s tant Lagged Acreage of Durum Wheat Total Annual Rainfall Relative Lagged Prices /PB | 550.91 + 0.41998 224.859769 0.24124 2.4500 1.74085. 360.57 + 0.41814 256.4816 0.232471 1.4058 1.3869.. 0.98073 0.707127 1.7986.. - 27.361 + 0.41720 1537.51 0.243049 + + 1.0074 0.74651 +453.54 1769.99 -196.01 1964.150 0.281428 0.789048 3.03 0.2016 2.03 0. 3204 2. 18 0.3248 -0.01779 1.71654. . 1.34947. . 0.25623 NS \^l t-1 Constant Lagged Acreage Total Annual Rainfall Relative Lagged Prices Relative Lagged Yields /YB ^ Notation: + 0.39944 + 1.0181 +723.61 - 24.044 VYD/t-l Significant at the 80% level -0.09979 1.41933.. 1.29027. . 2578.136 0.28067 NS 158.624 -0. 15157 NS NS Not significant 0.3266 ~4 76 The estimated acreage supply response functions are summarized in the table on the following page. The estimated coefficients of the lagged acreage and rainfall are positive and significant, their standard errors small, and the explanatory power of these two independent variables is significant. The estimated coefficient of the lagged relative prices (even though statistically insignificant) is negative indicating the recent tendency of Tunisian farmers to allocate less of their wheat acreage to Durum wheat when the relative price of this commodity is expected to be low. The high Durbin-Watson statistic indicates the absence of autocorrelation in the data used. Durum Wheat Yield Response The Traditional Production Period Durum wheat yield response model data covering the period (1949-1963). 1. 20 was fitted to time series The data used are: The amounts of fertilizer, computed as the ratio of the total quantity of nitrogen units used by the Tunisian farmers each year for Durum wheat production to the total acreage harvested at the end of the year. They are expressed in units of nitrogen per hectare; 20 This model was discussed earlier in Chapter II of this study. Appendix Table 8. Variables Regression The estimated Durum wheat acreage supply response in the modern production period. ^_ Coefficient Parameter S.E. Durbin-Watson R' Constant Lagged Acreage of Durum Wheat Constant Lagged Acreage of Durum Wheat Total Annual Rainfall Constant Lagged Acreage Total Annual Rainfall Relative Lagged Yields Constant Lagged Acreage Total Annual Rainfall Relative Lagged Yields Relative Lagged Prices 1 402.14 + 0.51594 182.876 0.21022 2.1989 2.4542* 180.51 + 0.51875 190.736 0.18533 0.94638 2.79904* 0.42117 2.17438. 205.281 0. 188109 0.43144 0.53659 2.6179* 1.93365. + 0.91579 110.15 + 0.49246 + 0.8 3426 + 77.475 786.96 + 0.43604 + 0.97662 + 98.965 -818.31 Significant at the 95% level. Significant at the 90% level. . . Significant at the 80% level. NS Not significant. 81. 4415 1031.518 0.210560 0.49098 89.5003 1220.906 0.951302 NS 1.92 0.3166 2.26 0.5097 0. 547 0.76291 2.07085. 1.98910. . 1. 10575 NS -0. 67024 NS 0.566 Notation: * -vl ^3 78 2„ The acreage of Durum wheat as reported by the appropriate services in the Department of Agriculture, and expressed in 1000 hectares; 3„ The National total annual rainfall as reported by the World Weather report at the end of each year. The estimated yield response equations are summarized in the table on the following page. The estimated coefficients of rainfall are positive and significant, their standard errors small, and the explanatory power of rainfall is significant. The estimated negative coefficients of acreage reflect the fact that Tunisian farmers were allocating lower productivity lands 21 to wheat production. The insignificant response of the local varieties grown on these lands to fertilizer application is reflected in a negative coefficient of fertilizer. The high Durbin-Watson statistic indicates the absence of autocorrelation in the data used. The Modern Production Period The yield response model was fitted to time series data covering the modern production period (1960-1974). The data used included the same items mentioned in the yield response model fitted to the traditional production period. 21 The number of observations is 15. More than two-thirds of the total acreage of wheat in the country are composed of marginal and poorly structured lands. Appendix Table 9. The estimated Durum wheat yield response in the traditional production period. Variables Regression Parameter 1 Constant Total Annual Rainfall 2.5576 +0.0056 0.84259 0.004103 3.03537 1.37769.. Constant Total Annual Rainfall Acreage of D. W. 4.5998 +0.008375 1.26671 0.003911 3.63125 2.141510* -0.00275 0.001374 -2.0050* Constant total Annual Rainfall Acreage of D. W. Fertilizer 4.6984 +0.00870 1.37254 0.004260 3.4231 2.0425* -0.00287 -0.02514 0.001499 0.09255 2 3 Notation: * Coefficient Significant at the 95% level. Significant at the 90% level. . . Significant at the 80% level. NS Not significant. 1 D. W. = Durum wheat. S. E. t -1.91618. -0.27163 NS Durbin-Watson R 2 1.43 0.136 1.84 0.367 0.372 80 Three regression equations have been estimated on a stepwise basis. The estimated yield response equations are summarized in the table on the following page. The estimated coefficients of rainfall are positive and significant, their standard errors quite small, and the explanatory power of rainfall is significant. The estimated coefficients of fertilizer are positive and significant, indicating a higher response of the new Durum wheat varieties, recently introduced in the country, relative to the local varieties. The negative coefficient of acreage, however, indicates that the quality of the land devoted to wheat production in the country, is still an important factor, although much less so than was the case during the traditional period. The absence of autocorrelation is re- flected in a high Durbin-Watson statistic. Appendix Tabl e 10. Variables Regression The es timated Durum wheat yie:ld re sponse in the modern prodAuction pieriod. Parameter Co efficient S. E. t Constant Total Annual Rainfall 1.013424 +0.013424 0.863507 0.00343857 1.74437 3.90395** Constant Total Annual Rainfall Fertilizer 1.3972 +0.01068 0.7892845 0.0034480 1.770227 3.098297** +0.076441 0.0400549 1.908398. Constant Total Annual Rainfall Fertilizer Acreage of Durum Wheat 2.2686 +0.012260 1.38025 0.0040525 1.643607 3.02517** +0.067247 -0.0013737 0.042423 0.00177008 1.585154. . -0.77604 NS Durbin- Watson R2 0.539 1.812 0. 646 0.665 Notation: ** Highly Significant (at the 99% level). Significant at the 90% level. . . Significant at the 80% level. NS Not significant. oo