Moroccan farmers have been cultivating durum wheat for centuries. In the 1980s, farmers began to switch from their traditional durum
varieties to modern varieties released by the Institut National de la
Recherche Agronomique (INRA) in collaboration with ICARDA.
A dynamic and efficient transfer of technology program (germplasm and crop management practices), initiated through the Mid-America International Agricultural Consortium (MIAC) project, played a key role in facilitating farmers' access to the new durum varieties.

Durum varieties released

By 1998, 49 durum varieties had been released in Morocco, 26 by INRA and the rest introduced from abroad, mainly from Europe, by private companies. INRA released 19 durum varieties between 1982 and 1998 (excluding private sector introductions and old varieties released prior to the establishment of the official catalog but listed as released in 1982). Widespread adoption cannot be achieved, however, without efficient seed production and distribution systems that ensure timely availability and adequate seed quality. But studies in the region have revealed an acute lack of synchronization between varietal release and seed multiplication and distribution of new developed varieties.

Contribution of the joint CIMMYT/ICARDA durum program

The durum improvement program of Morocco has benefited greatly from its close collaboration with the joint CIMMYT (International Maize and Wheat, improvement Center Mexico)/ICARDA wheat program based at ICARDA, as have other national agricultural research systems (NARS) in West Asia and North Africa (WANA). The origin of parent materials of varieties released by INRA provides a good indication of the contribution of CIMMYT/ ICARDA. With the exception of the very old varieties, e.g., 'Oued Zenati', 'Selbera', and 'Zeramek', virtually all durum varieties released in Morocco were derived from CIMMYT/ ICARDA-based material. Over the period 1988-1993, 72% of durum varieties released in Morocco were derived from CIMMYT/ICARDA crosses, while 28% had a CIMMYT/ICARDA parent. The reverse was true in 1994-1999, reflecting the improved capacity of the Moroccan program for crop improvement as a result of human resource development efforts under taken jointly with ICARDA.

Assessing the economic impact of durum improvement research

As with any other investment, agricultural research has costs and benefits. Among the criteria used to compare the profitability of investment decisions, the internal rate of return (IRR) is by far the most frequently used. It is the rate of return an investor expects to earn on a project over time, expressed as an annual percentage rate. The benefits of durum improvement research in Morocco were estimated using a simplified approach, based on the approximation of economic surplus, i.e., the change in the value of production resulting from the cultivation of modern durum varieties.
    Before calculating the returns to durum research, it is necessary to specify the time horizon over which the flows of expenditures and benefits occur and estimate the parameters that enter into the calculations of durum research-induced returns, e.g., adoption rate of modern durum varieties, genetic yield gain of modern varieties, and durum prices.

Time horizon:
   
The impact assessment spans 1970-1999, demarcated into two sub-periods: 1970-1982, corresponding to the development, on-farm testing, release, and early diffusion of the first modern durum varieties released by the Moroccan national program, and 1983-1999, corresponding to the increasing adoption of these first varieties and those released after 1983. Therefore, 1970 is set as the starting point of the flow of durum research expenditures (ceasing in 1995). Returns, on the other hand, consist of a continuous annual flow of benefits starting in 1983 and ceasing in 1999. This implies a time lag of 13 years between the initiation of research and the time farmers have access to released varieties (9 years to release the variety and 4 years for seed production and extension). The length of expenditure and benefit streams were set in this manner to arrive at a "least returns" scenario of the durum improvement program.

Adoption of modern durum varieties:
 
The rate of adoption is a key determinant of the size of returns to crop improvement research. The evidence of high rates of adoption of modern durum varieties is well documented in Morocco. Rates of nearly 80% and over 90% for durum and bread wheat, respectively, have been reported. Results of a 1997 survey showed that 98.2% of durum area of sample farmers was planted to modern varieties, even in arid zones. However, out of the 19 durum varieties released between 1982 and 1997, only seven were grown in 1997 by sample farmers. Two varieties, 'Marzak' released in 1984 and 'Karim' released in 1985, were found to occupy three quarters of the total area planted to modern varieties. In 1997, the newest varieties grown (each occupying less than 5% of total area under modern varieties) were all released in 1988, implying a rate of varietal replacement of about 12 years, as indicated by the weighted average age of durum varieties (Table). Given this long time lag, farmers will not fully benefit from the desirable traits incorporated in the new varieties. This delay in flow of benefits lowers the calculated returns of durum research. For the purpose of this study, it was assumed that peak adoption, conservatively set at 75%, is reached after 12 years.

Genetic yield gain of modern varieties:
   
Benefits from crop improvement can take the form of higher yields, improved yield stability, better grain/straw quality, lower production costs, etc. A series of four-year on-farm yield trials conducted between 1982 and 1998 showed that modern varieties have a clear productivity edge (especially the so-called second generation modern varieties) over traditional varieties. But the genetic yield gain of modern varieties is not sufficient, by itself, to generate benefits. Modern varieties can have a positive impact on crop productivity at the farm level only if there is sufficient supply of seed and if farmers adopt them. Widespread diffusion occurs only when farmers are convinced that modern varieties hold a clear advantage, in terms of cost and benefit. The impact of modern durum varieties on the growth of durum productivity was assessed using a yield time series covering 1970-1999. The results indicate a significant. acceleration in the rate of durum yield growth in the period 1983-1999 (2.8% annually), which corresponds to the spread of modern varieties. Although this estimated annual rate of yield gain due to modern varieties might seem high, similar rates were obtained in other studies in Nepal (2.5%) and Morocco (3.8%). The relatively high acceleration rate of yield growth estimated most likely resulted from the combined effect of modern varieties, input intensification, especially chemical fertilizers, and increased use of supplemental irrigation on areas planted to modern varieties. Therefore, the net modern varieties-induced rate of yield gain used to derive the benefits of durum improvement research in Morocco was conservatively estimated at 1.40% per annum, i.e., 50% of the calculated acceleration rate.

Durum prices:
 
Because in Morocco durum wheat prices are not controlled, no durum producer price series is available. In general, regardless of the year, durum prices are at least 15% higher than bread wheat prices. However, FAO-published bread wheat prices (1985-1995) in Morocco are much higher than durum CIF (cost, insurance, and freight) reference import prices, implying that actual durum prices in Morocco would not likely provide a good approximation of durum economic prices and would inflate the internal rate of return. Therefore, CIF prices converted to Moroccan currency, using an average exchange rate of 10 dirhams (Dhs) to the dollar, were used instead.
    These parameters were then used to calculate the stream of gross benefits resulting from the gain in durum production attributable to durum research.

Durum research expenditures:
   
The stream of benefits must be compared to the stream of costs associated with durum research in order to determine the stream of net returns. The estimation of annual durum research expenditures was done in two steps. First, the number of scientists (full-time equivalents [FTE]) involved in durum research from 1970 to 1995 was determined in collaboration with INRA. Second, research expenditure data were gathered from INRA. These data included salaries and fringe benefits of scientists, including MIAC expatriate salaries, and associated support staff, research management costs of experiment stations where research was conducted, direct administration costs, training costs, funds allocated to durum research through projects involving CIMMYT, ICARDA, and other regional and international institutions, and overheads. These costs were then used to calculate expenditures per durum scientist. Moreover, an extension cost, estimated at 15% of total staff salary, was added. Using this approach, the 1995 expenditure per researcher was estimated at 1,500,000 Dhs, i.e., around US$150,000. (This expenditure per scientist includes costs that are usually not considered in most research impact studies, and is considered an acceptable approximation of research costs per durum scientist in Morocco.) Nominal expenditures per year were then obtained by multiplying this expenditure per researcher by the number of scientists (in FTE) for each year from 1970 to 1995.

Economic impact of durum research:
   
Based on the calculated benefits and costs, the IRR for durum improvement research for the period 1970-1999 was estimated at 32%. As indicated, this IRR not only accounts for returns to durum breeding but also to breeding support activities (pathology, entomology, physiology), crop management (agronomy), as well as foundation seed production. The estimated IRR for durum improvement research in Morocco falls in the lower side of the range of reported IRRs obtained in Morocco and other WANA countries for other cereal crops, e.g., bread wheat and barley. Recently, for example, an IRR of 69% for barley breeding was reported in Morocco, despite relatively low adoption rates (maximum 19%) as compared to the relatively high adoption rates reported for durum. An IRR of 32% implies that if the national program of Morocco had to borrow the funds invested in durum improvement research between 1970 and 1995 at an interest rate of 32%, the benefits generated by durum research over the 1983-1999 period would be sufficient to repay the principal and accrued interest on such a loan. The net present value of the stream of expenditures and returns over the 1970-1999 period amounts to about US$142 million. Therefore, although the estimated IRR might not compare favorably with previous rates reported in the region and elsewhere, it nevertheless represents a good return on investment.

    Clearly, reported IRRs reflect the assumptions used in evaluating the research investment considered. The estimation of returns to agricultural research is obviously sensitive to the parameters generally incorporated in the estimation of gross annual research benefits. Moreover, in addition to the size of research costs and benefits, their distribution through time is of paramount importance. Because of discounting, IRR will greatly depend on the lag between the first year that research expenditures are incurred and the first year that research benefits flow. For durum research in Morocco, this lag is set at 13 years. Moreover, for the first three years of benefits, adoption is usually so low that benefits are negligible. It is clear that because of discounting benefits accruing toward the end of the period of analysis, which coincides with relatively large gains in production, a much lower present value is arrived at.

On the other hand, however, costs incurred early in the research process will have a higher present value.
Sensitivity analysis clearly shows that substantial increases in returns to investment in durum research could be achieved through the reduction of the research lag. For example, reducing the research lag to 9 years, instead of 13 years, would lead to a 21% increase in IRR, while a reduction of the research lag to 5 years would lead to a 40% increase in IRR. As a matter of strategy, the Moroccan durum research program should set the reduction of the research lag as a priority, in view of the potential economic implications.

Conclusion

Although derived under a rather conservative scenario, the IRR obtained provides strong support for sustained investment in durum improvement research in Morocco. Policymakers must understand that in order to prevent extreme fluctuations in production, such as those experienced in 1994 and 1995, when, in the favorable zone, average durum yield fell from nearly 2 t/ha to less than 0.4 t/ha, more research is needed in breeding (for resistance/tolerance to drought) and crop management to enable farmers to use soil moisture as efficiently as possible (soil moisture is the binding constraint to crop productivity under rainfed conditions.) Durum improvement research, an important component of INRA's portfolio, should be strengthened for at least two reasons: 1) the development and utilization of modern durum varieties with desired traits is a cost-effective way to increase durum production, and 2) unlike knowledge-intensive technologies, early and widespread use of new modern varieties by farmers, including smallholders, can be rapidly and substantialy improved. In addition to economic considerations, increased investment in durum research is also justified on food security and equity grounds, given that it reduces the vulnerability of smallholders to fluctuations. Moreover, as evidenced by this study, durum improvement research generates relatively high returns. Therefore, any policy decision associated with durum research will necessarily be of high significance.