信息经济学-不确定性、风险与信息.ppt

* Pooled data - mix of both elements e.g. profit rates for firms over a ten year period. e.g. profit (it) = f(conc, capital/labor ratio) Panel data - follow a household over time. Data can be quantitative - prices, income, money supply; or qualitative - male/female, married/unmarried. Lots of problems with data - may be variables you would like to get but cant, may be rounding errors, may be low response in surveys. Results of research are only as good as the data you use. But have to try. Always state sources of data used, definitions, limitations, assumptions used, so your results can be interpreted correctly. Need to look at probability and statistics because in estimating econometric models, we are concerned with the statistical relationships among variables, not deterministic as in say classical phsyics - law of gravity an exact relationship - F=k((mass1*mass2)/distance squared). In looking at statistical relationships we deal with random variables, or stochastic variables - variables that have probability distributions. (In fact, the dependent variable is assumed to be random, to have a probability distribution, but the independent variables are assumed to be fixed, have fixed values in repeated sampling - even though they may be intrinsically stochastic; note random=stochastic=probabilistic). Note also that we look at the dependence of one variable on another, but this does not necessarily mean one variable causes another - although it may. We have to look at theory to figure out whether or not we can reasonably assume causation. (Take Philips curve - do high money wages lead to lower unemployment, does lower unemployment lead to high money wages, or does some other factor cause both?) * Pooled data - mix of both elements e.g. profit rates for firms over a ten year period. e.g. profit (it) = f(conc, capital/labor ratio) Panel data - follow a household over time. Data can be quantitative - prices, income, money supply; or qualitative - male/female, ma