An International Comparison of the Effects of Government Agricultural Support on Food Budget Shares

Journal of Agricultural and Applied Economics, Aug 2008 by Miller, J Corey, Coble, Keith H

Model

We use the SAS software package to execute a time series cross-sectional regression model according to the Parks method, which provides for an autoregressive model. Table 1 presents the results of this model. For the model in Table 1, R^sup 2^ = .434.

Results

The independent variable for value added to GDP by agriculture is not significant at the .05 level, indicating the relative size of each country's agriculture industry does not significantly affect our measure of food affordability. The average value as a percentage of GDP does not exceed 10% for any of the OECD countries we examined, and the values change relatively little for most of the countries over the 1986-2004 period.

The agricultural productivity index is statistically significant and negative, indicating that as a country's agricultural technology improves, food becomes more affordable by our measure. This expected finding concurs with the results of the models of Miller and Coble.

PSEs as a percentage of agricultural receipts is also a significant and negative variable at the .05 level. This variable indicates that as a country's PSE estimate becomes larger relative to agricultural receipts, food by our measure becomes more affordable. Given the levels of PSEs provided by several of the OECD countries in our model, as noted by Huffman et al. and others, this finding is not entirely unexpected.

The other variable measuring support to agriculture, the CNPC, is also significant. Its coefficient value is positive, however, indicating that protectionist measures make food relatively more expensive for consumers by our ratio variable.

The remaining variables reported in Table 1 represent the two sets of dummy variables for the group of OECD countries with relatively high support and protection measures (Iceland, Japan, Norway, and Switzerland) and those with relatively low support and protection measures (Australia, New Zealand, and the United States). Because of the nature of the panel regression model used, the significance and sign of the coefficients on these variables, one should interpret them relative to the corresponding independent variables.

Both dummy variables for valued added by agriculture lack significance, indicating that neither is significantly different from the value-added independent variable. Thus, because this variable is not significant, value added by agriculture lacks significance for any of the OECD countries in our model.

The dummy variables for agricultural productivity are both significant and positive for the "high" and "low" support countries. These coefficients indicate that, for both groups of countries, agricultural productivity has a smaller effect on our ratio variable relative to the effect on Canada, Mexico, and Turkey. The reason for this finding is not entirely clear, although the agriculture productivity indices for Japan, Norway, and Switzerland all trend down over the 1986-2004 period.

The dummy variable for PSE as a percentage of agricultural receipts for the "high"support countries is not statistically significant, indicating that the effect of PSEs for these countries is similar to that for Canada, Mexico, and Turkey. The relatively high levels of support to agriculture provided by Iceland, Japan, Norway, and Switzerland might explain this finding. The coefficient on the dummy variable for PSEs for the "low"-support countries is significant, positive, and roughly the same size as the coefficient for the independent variable for PSEs. The finding that these two PSE variables could effectively offset each other indicates that the support provided to agricultural producers in Australia, New Zealand, and the United States has little effect on our ratio variable. Such a result is consistent with those of Miller and Coble and, given that Australia and New Zealand provide the lowest levels of support among OECD countries, makes sense.