Spriggs, 1991, Zanias 1998, Saghaian, Reed and Marchant 2002). However, Bessler (1984),
Grennes and Lapp (1986) Robertson and Orden (1990), and Cho et al. (2004) found that
relative agricultural prices are not affected by nominal macroeconomic variables. These
studies also show that although short run effects of money changes may be different, long run
effect are equal supporting the long-run neutrality of money (Ardeni and Rausser 1995).
However, Saghaian et al. (2002) results reject the hypothesis of the long-run neutrality of
money. It should be noted that these results should be interpreted only with care. First, time-
series studies of links between the agriculture and the rest of economy are often sensitive to
variable choices. Second, as Ardeni and Freebairn (2002) pointed out, many studies lack an
appropriate treatment of the time series properties of data implying misleading results
especially on the case of earlier research. Finally, the main feature of the literature is that
many studies do not relate directly a specific macroeconomic model, except Saghaian et al.
(2002), rather they use a set of explanatory variables suggested by previous studies.
3. Empirical Procedure
Even as many individual time series contain stochastic trends (i.e. they are not stationary at
levels), many of them tend to move together on long run, suggesting the existence of a long
run equilibrium relationship. Two or more non-stationary variables are cointegrated if there
exists one or more linear combinations of the variables that are stationary. That implies that
the stochastic trends of the variables are linked over time, moving towards the same long-term
equilibrium.
3.1. Testing for unit roots
Consider the first order autoregressive process, AR(1):
yt = ρyt-1 + et t =...,-1,0,1,2,..., where et is White Noise. (1)
The process is considered stationary, if I ρ∣ < 1, thus testing for stationarity is equivalent with
testing for unit roots (ρ= 1).
(1) is rewritten to obtain
∆yt = δyt-1 + et , where δ = 1 - ρ (2)
and thus the test becomes:
H0 : δ = 0 against the alternative H1: δ < 0.