Which model should I use for Time series data? I need to regress one dependent variable (dummy variable), against several other independent variables (dummy and non dummy variables). (FYI : I'm not using the past performance of dependent variable, the independent variables are different.) I have time series data for loans. My dependent variable is default status (1, if defaulted , 0 otherwise) and independent variables are loan characteristics. I know I need to check for stationarity among the dependent and independent variables. But what is the next step? Can I use classical linear regression model after I check for stationarity (and the data is stationary)? Is there a specific model to use in this case. Can I then just follow up with the residual diagnostic tests and stability diagnostic test (Ramsey's reset)? Is this sufficient? If not, can someone explain what is step by step process in layman's terms. I am quite new to econometrics and would really appreciate any help. Thank you

@InstitutoEconometriaLima Thank you so much for your response. No the dataset is the loanbook of one institution for 2 years. It includes a balanced sample of all loans provided in this period. I considered it as time series data as the variables are all of one institution. Although I do not use lagged values of Y (dependent variables) as the explanatory variables, the multiple independent variables are all derived from the same dataset. My dependent variable is default status of loans and independent variables are loan characeteristics. M independent variables include dummy variable and quantitative variables. I have about 6 to 7 independent variables. All of the dummy independent variables are non binary, as I have more than 4 categories and even 15 in some case. For example, one of the independent variable is country and the sample has 15 countries. I made n-1 dummy variables, i.e. 14 dummies and coded them as 1 if it is country and 0 otherwise. I left one variable with highest number of observations as the base category. I also have another dummy variable as purpose of loan and includes 5 categories, I used the same approach. I actually used logit regression, however, I get "quasi separate completion error" and says one variable of the purpose of the loan completely predicts the relationship. I dont think it is true as this is not the main independent variable in the study, I used it as a control variable. I have already filtered defaulted loans from excel and saw that all loans have the independent variable characteristic present. I dont know if I face this error because there are too many dummy variables when I estimate the equation. Could you advise? Also, I have another hypothesis in the same study, only the dependent variable is different and is in percentage. I tried using classical linear regression model with OLS method of estimation via E-views, since my data is stationary. However, even that results in perfectly collinear error (I dont remember the exact name of the error). I would really appreciate if you could provide your email if you are willing to help further. Thank you again.
 A: Richard. I hope this helps.

*

*If it's indeed a time series data, it's a single loan or an aggregated loan, isn't it? Those characteristics have to be attributed to that single/aggregated loan as well. If it's not the case, you may have multiple loaners whose characteristics vary along the time, which suggest you have a (dynamic) panel data. Please clarify.

*As for the model, if it's time series with a binary dependent variable, the general solution is to use a Logit model (Logistic).

*Stationarity is recommended but if your variables aren't macro, they might not be non-stationary (sorry double negative).

*After the estimation, you can use similar residual diagnostics, but in addition sensitivity/specificity, ROC tests (among other) related to logistic models. (i'm assuming no lags on dependent variable).

*Ramsey's test could be of help, but remember it's for omitted variables only.

*If it's not an autoregressive model/dynamic (actually, the main advantage of using times series techniques) then modern cointegration could be off topic (However, it's not usual). A classical cointegration Engle-Granger test could be sufficient.

