3 A review of previous studies on dairy farm efficiency and productivity
3.2 Previous studies in dairy farm efficiency
3.2.1 The underlying assumption about technology
According to the frontier approach, it is assumed that all farms are confronted with a single production frontier and therefore share the same underlying technology. Differences among farms arise in the efficiency with which technology is used. Assuming farms share the same technology when they do not will result in biased measures of efficiency and confusion among technological differences.
Few studies on efficiency addressed the issue of technological differences among farms. One early example is Kumbhakar, Biswas and Bailey (1989). They divided the sample of Utah dairy farmers into three groups according to total value of sales. They then estimated the frontier of each group and the pooled frontier (all groups together). The likelihood- ratio test (LR test) was used to test the null hypothesis that all farms operated under the same technology. They accepted the alternative hypothesis that production structure differs across farms of different sizes (p 601, footnote 4). Similarly, Hallam and Machado (1996) tested whether specialised6 and non-specialised Portuguese dairy farms operated under the
same technology. Hypothesis testing was performed by a Chow test for equality of parameters of frontiers estimated for both farm types. They accepted the null hypothesis that both sub-samples are confronted with the same production frontier.
Mbaga, Romain, Larue and Lebel (2003) divided the sample of Quebec dairy farmers into two groups (non-maize and maize regions) to assure homogeneity of exogenous conditions. They then estimated the frontier for each subsample. However, they did not test whether technologies were different even when some results pointed in that direction. In practice, farms may adopt different technologies for a variety of reasons. Soils, climate and landscape differ across regions, influencing for example, the amount and type of feed grown, the opportunity cost of land and the level of scale economies (Sumner and Wolf, 2002). In turn, these exogenous conditions may impose some restriction in the selection and type of technology used. Thurow and Holt (1997) stressed the importance of heritage and past investments in determining the type of technology that is most successful, as well as the management preferences of individual farmers.
The evolutionary theory claims that technologies evolve along specific pathways or trajectories (Brennan and Wegener, 2003). Innovation is usually a continuous incremental process within a technological regime. Rather infrequently, innovation is radical, i.e., abandonment of a particular technological regime.
Arthur (1989) asserted that a dominant technology could be progressively “locked-in,” seriously restricting the movements from one state to the next and confining innovation to a narrow corridor of developments. Adoption of a new capital good may require changes in existing equipment, bringing in additional cost of adjustment. In addition, existing management and labour skills may limit or even prevent innovations. This “interrelatedness” limits the scope of adoption (Brennan and Wegener, 2003).
Other factors, like uncertainty about the performance of new, unproven technology, coupled with the irreversibility of some investments in fixed assets may restrict adoption decisions (Purvis, Boggess, Moss and Holt, 1995). Related to irreversibility, there is also the issue of the non-transferability to other uses (e.g., a milking machine can only be used for dairying).
Therefore, at the regional level, innovation is incremental and strongly shaped by existing socioeconomic structures and the behaviour of their agents. Radical innovations, in turn, tend to appear in new areas and are less predetermined and dominated by successful
structures (Tödtling, 1992). Similarly, at any point in time, new firms entering the industry are confronted with different technologies. The choice of a particular technological regime is largely random and only through ex-post competition, can uncertainty about competing design be resolved (Arthur, 1989). Following the logic of the evolutionary theory, some educated guesses can be made about the outcomes of the choices the new firms face. New firms entering the industry in a well-established region will be prone to adopt the dominant paradigm. On the other hand, if the new entrant chooses a new location where a priori there is no dominant technological paradigm, it will be less conditioned and more able to adopt a different technology (Tödtling, 1992).
The above-mentioned is particularly true in agricultural production, where the interaction of geophysical factors (location-specific) and the adopted technology may result in different outcomes. As Alston (2002) asserted, the biological nature of agricultural production implies that the spatial dimension is significant for agricultural technology. In fact, a successful technology applied in one location may not be entirely transferable to a new location.
Given the spatial dimension of agricultural technologies, it is important when measuring efficiency to correctly identify the technology applied. Otherwise, inefficiency will be confused with using an inferior technology (Battese el al., 2004). Fraser and Cordina (1999) stressed the importance of assuring homogeneity of exogenous conditions (soils, climate and physical parameters) likely to affect efficiency. They asserted that such technical efficiency differences are the results of managerial ability. Presumably, they were stressing the importance of ensuring that all farms operate under the same technology in order to gauge the correct efficiency estimates.
In view of the above-mentioned, it would be more accurate to say that assuring homogeneity of exogenous conditions (soils, climate and physical parameters) is crucial insofar as technology applied is likely to be the same (albeit with different rates of adoption) and hence true estimates of (in)efficiency may be obtained. Technical efficiency differences from a group of farms that share homogeneous exogenous conditions and operate under the same technology are, therefore, the result of managerial ability.
The geographical spread of the datasets of the papers reviewed is mixed. Only one study has a district scope (Fraser and Cordina, 1999). Six of them are based on provincial or state data and another six on regional data. The rest are classified as follows: one deals with farms in the province of Ontario, Canada and the state of New York, United States (Haghiri, Nolan and Tran, 2004), one pooled data from two countries (Dawson and White, 1990), while nine have national/country coverage.
Bravo-Ureta and Rieger (1991) attempted to correct their efficiency estimates by introducing location dummies to capture effects on the placement of the production technology. However, the coefficient of the slope parameters in the production function were the same for all farms, i.e., all farms face the same frontier.
Haghiri, Nolan and Tran (2004) pointed out that the selection of the province and state for the inter-country comparison was done on purpose, given the similarities in production technology and geophysical conditions between them. Even though the non-parametric stochastic frontier was estimated for both samples independently, no formal test was conducted to check whether technology applied was the same.
Regarding the studies at a country level, Piesse, Thirtle and Turk (1996) mentioned that most of the sample farms have similar alpine terrain, which assures homogeneity of exogenous conditions. Heshmati and Kumbhakar (1994), Kumbhakar, Ghosh and McGuckin (1991) and Kumbhakar and Heshmati (1995) introduced regional and size dummies to accommodate possible differences in productivity among different regions and size, but all farms face the same frontier. Kumbhakar and Hjalmarsson (1993) explicitly account for farm-specific characteristics related to location, climate and land quality, given that inefficiency should not be confounded with farm-specific characteristics.
For the NZ studies (Jaforullah and Devlin, 1996 and Jaforullah and Whiteman, 1999), the authors assumed that all farms face the same frontier and hence, all farms applied the same technology.