Demographic influences on willingness to pay for cold tolerance technology.

The survey responses showed that there are more farmers in older age categories, as shown in figure 1. The survey sample contained older farm operators than would be representative of Alberta. According to the census, one-half of all farmers fall between the ages of 35 and 54 years (Statistics Canada 2001). In this sample, nearly half of all respondents were over 51 years of age. This may have impacted the results of our analysis, as age may be related to adoption behaviors and attitudes toward technology. One explanation for the difference between the census and our data could be that older farmers are more likely to attend the extension events at which the survey was administered.

[FIGURE 2 OMITTED]

The average farm income level of respondents to the survey differed greatly from the Alberta data, as displayed in figure 2. While most farmers reported a net income of less than $25,000 on the census, over one-half of cold tolerance survey respondents placed themselves in the over $50,000 annual net income category (Statistics Canada 2001). This could be because farm operators with more income are more likely to attend extension events.

The mean education level of survey respondents was college or technical school. The full distribution of responses is shown in figure 3. The survey sample contained people with a higher average level of educational attainment than the general farm population In particular, there was a higher proportion of university-educated people in the sample (Statistics Canada 2001).

[FIGURE 3 OMITTED]

Figure 4 illustrates that the farm sizes reported by respondents to the survey were larger than the provincial average (Statistics Canada 2001). This could be because the survey focused on large-scale grain farmers rather than smaller livestock farms or hobby farms. Small farmers often have off-farm jobs, and this may have decreased the likelihood of attendance at events where the survey was performed.

Past Behavior and Experiences

The pie chart in figure 5 shows that frost damage is a consideration for the vast majority of producers in the sample. A closer examination of the data reveals that the majority of frost damage is experienced late in the season.

These numbers indicate that crop damage due to low temperatures is a commonly experienced problem, so there should be a market for more cold-tolerant varieties. At the most basic level of analysis, the dichotomous choice variety questions revealed that 39% of respondents indicated that they would be interested in a new variety with some type of improvement in frost-tolerance characteristics, even though this came along with a seed price increase (see figure 6).

[FIGURE 4 OMITTED]

Logistic-Stated Choice Results

The regression, detailed in table 1, was developed using TSP software. Demographic factors were used as independent variables in the regression. A logit model was used, which allows for both choice-specific and chooser-specific variables to be used as explanatory variables (Hull and Cummins 2005).

A result of one indicates a decision to stay with the existing variety, while a result of two indicates a decision to invest in a new, more cold-tolerant variety. The equation form is as indicated below. The numbers to be used in this formula are in table 2. Probability (choose new variety) = [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII], where [X.sub.i] = variety or respondent characteristic and [[alpha].sub.i] = estimated logit coefficients. The first variables in the equation include: SC, FT, DD, and a status quo or constant variable (SQ). These were the choice specific variables, which varied depending on the survey version and the new variety being examined. Demographic (chooser-specific) variables included were: farm size in acres (SIZEA2), age in years (AGE2), education level (EDUCATION2), income level (INCOME2), region (REGION2), and frequency of frost damage experience (HOWOFTEN2). These demographics were not interacted with the seed characteristics before addition to the model. Examining the coefficient on each of these variables gives insight into the type of people more likely to invest in cold tolerant cereal seed, in this case the new variety. The scaled r-squared of 0.180615 indicates that this model has just over 18% more explanatory power than a model containing only a constant term (Hall and Cummins 2005).

The coefficient on SC is negative, indicating that utility gained from the new variety drops as the seed price increases. This follows the general law of demand and was not a surprising result. Both increased degrees of frost tolerance and decreased days to maturity had positive coefficients, indicating that the likelihood of purchasing a new variety increases as tolerance to cold temperatures increases or as the variety matures faster. These things both increase the risk reduction value of the product, so they increase the value of the seed and therefore propensity to adopt and willingness to pay. Each of these first three variables in table 2, SC, FC, and DD, is statistically significant at the 5% level.

Farm size has a statistically significant positive coefficient. Operators of larger farms are more likely to adopt the new varieties. This could be because large farms are faced with more risk that they would like to mitigate because they are less diversified, often the operator's only source of income, or have a larger budget for investment in new technologies. This follows the findings of Karshenas and Stoneman (1993), but it is in contradiction to the findings of Koundouri, Nauges and Tzouvelekas (2006). The introduction of a new wheat variety does not require the labor and management resources demanded by the irrigation systems examined in the Kondouri, Nauges, and Tzouvelekas study, so this could explain the differing results. Larger farms can adopt more cold-tolerant seed without adding extra work to their operation.

There was a negative coefficient on the age variable, but this coefficient was not statistically significant at the 5% level. The negative sign follows in line with findings in other studies, such as one by Saha, Love, and Schwart (1994), where younger operators were more likely to try new things, but the data from this project does not strongly support this conclusion.

One of the results from this analysis that contradicts past findings is in the effects of education. The negative coefficient (table 2) indicates that as education level rises, the utility created by adoption and the likelihood of adoption decreases. Past work by Koundouri, Nauges and Tzouvelekas (2006) and many others indicates the opposite; in general those with more education are more apt to adopt new technology. The reasoning behind this finding is unclear, but it could be because those with varying education levels did not systematically differ from one another in other ways. Income, often closely related to education, did have the expected sign on its coefficient.

Region is one of the more interesting variables in this equation. The coding on the survey was as shown in the map in figure 7, with one being the furthest north region, the Peace. There are great variations in weather and temperature throughout Alberta, and there is a greater risk of frost in the more northern regions. Thus, the negative coefficient on the region variable is logical, as it means that those in the lower numbered areas are more interested in frost tolerance and decreased days to maturity as a result of their location. The Northwest, region 2, is more mountainous and forested than region 3, the Northeast, so this follows along the correct gradient. The coefficient on this variable was not significant at the 5% level, but it was significant at the 10% level, and seemed like an important finding although it had a fairly high p value. The map included in figure 6 indicates the regions and the percentage of respondents from each region. The remaining 3% of respondents were from other provinces or did not indicate the area in which they farm.

[FIGURE 6 OMITTED]

The variable measuring frequency of frost damage (HOWOFTEN2) follows a similar pattern to that shown in region. The categories for this variable were: (1) every year, (2) every 2-3 years, (3) Every 4-10 years, and (4) Less than every 10 years. There was a statistically significant negative coefficient on this variable. This indicates that respondents who experience regular frost damage are more likely to invest in more resilient varieties in order to minimize losses experienced on these occasions.

[FIGURE 7 OMITTED]

Willingness to Pay

After estimating the logit model, it was possible to calculate the willingness to pay for the variety improvement of greater FT and earlier maturity. First, values for a representative respondent were chosen based on mean demographic values from the survey. This respondent farmed 2,885 acres, was between 41 and 50 years of age, had a college or technical school education, and made $50,001-$100,000 per year in net income. The respondent was from the Northeast region and experienced frost damage every 4-10 years. It was determined that the willingness to pay for increased degrees of frost tolerance, decreased days to maturity, and a combination of the two were all just slightly negative in this case. From there, the focus shifted to those who experienced frost damage on a regular basis, as this will be the target market for these varieties.

[FIGURE 8 OMITTED]