How to organise data and run nerlovian model I am trying to estimate the price elasticity of supply for small scale farmers in Malawi and I have time series data for 34 years. I have two problems: First, the prices are very low such that when I take their logs am getting negative values, as a remedy i added 1 to each value and i got positive logs but i don't know how to interprete the result with the 1 that I added. Second i want to find out the specific way of estimating the model in Stata, I just used regress. I will appreciate your help
 A: The non-linear structural equations of the Nerlove model reduce to a simple linear equation:
\begin{equation}
\ln y_{it}=\beta_0+\beta_1 \ln y_{it-1}+\beta_2 \ln y_{it-2}+\beta_3 \ln p_{t-1}+\beta_4 \ln x_{it} + \varepsilon_{it},
\end{equation}
where $x$ is any exogenous non-price variables (like rainfall). The coefficient on lagged price gives you the short-run price elasticity of supply, and you can interpret the effect as saying that for 1% increase in price, you get $\beta_3$% increase in output. You definitely don't need to add the one, like Nick pointed out. Don't take log of binary variables.
With panel data, xtabond can be used, but the dynamic panel route demands a fairly high level of technical sophistication. Applied folks will often estimate this using OLS like you did, even though this is not entirely correct.  
A: Negative logarithms are not a problem. All they mean is that your prices are less than 1 in whatever currency you are using. This is basic mathematics: e.g. http://en.wikipedia.org/wiki/Logarithm
Thus you should not add 1 before taking logarithms. 
That said, I don't know how you get price elasticity out of time series because I am not an economist, but I suspect no harm would be done by your telling us more about your dataset, notably what variables you have. 
