Bayesian logistic regression model with 2 random effect intercepts not converging

Question

I am trying to model some data using a Bayesian logistic regression model. The data are monthly observations of a binary outcome event across multiple households. So I have about 400 households that were observed for 48 months and each month the outcome of the binary event was recorded (1/0). As a first step I tried to model the probability of the event using a Bayesian logistic regression model with just a random intercept for the household and the month. I used RJags to estimate the posterior probabilities and I am not having any success getting the chains to converge. I wonder if I am not specifying the model correctly?

Here is my model specification

modelString = "model {

### MODEL

for (n in 1:N){
delta[n] ~ dnorm(Mu_d, Tau_d)
}

for(t in 1:tt){
alpha[t] ~ dnorm(Nu_a, Tau_a)
}


for ( n in 1:N ) {
 
  
    for (t in 1:tt){
      
      logit(theta[n,t]) <- alpha [t] + delta [n]
                           
     
       y[n,t]~ dbern(theta[n,t])
     
Mu_d ~ dnorm(0,0.01)
Tau_d <- 1 / (sd_d * sd_d) 
sd_d ~ dt(0,1,1)T(0,)   ## Specify a half-Cauchy prior 

Nu_a ~ dnorm(0,0.01)
Tau_a <- 1 / (sd_a * sd_a) 
sd_a ~ dt(0,1,1)T(0,)   ## Specify a half-Cauchy prior 

 }"

The input data (ins) is a matrix with households along rows and months along columns.

Would appreciate your insights.

Thanks!