I have some macroeconomic variables. In annual series i have data set of 45 years and in monthly format, i only have 252 observations. I want to study their volatility. As far as standard time series volatility model, ARCH and GARCH model requires large samples (Ng & Lam and Hwang & Pereira 2006). Some empirical studies used elasticity approach to define volatility. Few studies uses OLS based model like following-

ln(Yi)= ai+bit+ei (Trend Linear Equation)

then extracted the error term and measures the standard deviation of it. In this equation, is the stationary problem has any implications ??? How can i measure volatility of these variables ??? What could be the alternatives ?? Can we extract any information form applying ARCH/GARCH models here ??