Help with loop code for a factor model

[pao.martos]

Hello everyone,

I have a dataset (nrow = 10,000 ncol = 29) called random_draws for a factor model.
the first 27 columns are exchange rate returns, 28th and 29th columns are the factors observations.

I want to calculate exchange rate returns with the following factor model formula:

FX_returns = mean(i) + std_returns(i) * (rho1 * w1 + sqrt(1 - rho1^2) * w(i))
+ std_returns(i) * (rho2 * w2 + sqrt(1 - rho2^2) * w(i))

w: factor values
w(i): idiosyncratic values

The code I have right now is the following:

set.seed(123) # Set a seed for reproducibility
num_draws <- 10000
num_pairs <- 27
num_variables <- num_pairs + 2

for (i in 1:num_pairs) {
rho1 <- factor1_loadings[i]
rho2 <- factor2_loadings[i]
factor1_shocks <- random_draws[, num_pairs + 1]
factor2_shocks <- random_draws[, num_pairs + 2]
independent_shocks <- random_draws[, i]
FX_returns[, i] <- mean_returns[i] + std_returns[i] * (rho1 * factor1_shocks + sqrt(1 - rho1^2) * (independent_shocks)+ std_returns[i] * (rho2 * factor2_shocks + sqrt(1 - rho2^2) * independent_shocks)
}

factor1_loadings is a vector of factor loadings for the first principal component I got from a PCA and factor2_loadings is the same but for the second principal component
factor1_shocks are observations from the 28th column in the dataset
factor2_shocks are the observations from the 29th column in the dataset
independent _shocks are the observations per column from 1-27

I have calculated the mean and standard deviation of each column 1-27 which are the mean and standard deviations for each exchange rate

I want to place the FX returns outputs in a dataset with nrow = 10000 and ncol = 27 (same number of exchange rates)

The error code I keep getting is this one:
Error in FX_returns[, i] <- mean_returns[i] + std_returns[i] * (rho1 * :
incorrect number of subscripts on matrix

Can anyone help with how to fix this code so that it works?

[technocrat]

@pao.martos

FX_returns[, i] <- mean_returns[i] + std_returns[i] * (rho1 * factor1_shocks + sqrt(1 - rho1^2) * (independent_shocks)+ std_returns[i] * (rho2 * factor2_shocks + sqrt(1 - rho2^2) * independent_shocks)
}

This is as bar as it's possible to follow the problem because of a syntax error. When corrected, the next problem is that factor1_loadings is not in namespace. These are problems that a reprex (see the FAQ can solve to get everyone on the same page.

So, I can only offer generic advice.

1. When the incorrect number of subscripts on matrix error goes get fixed, the for loop will not return anything because Fx_returns is created in the local environment of the function and is not returned to the global environment. And the usual attempt to make a return almost always returns just the final iteration. The usual fix is to initialize Fx_returns in the global environment here as a matrix object with dimensions equal to the number of rows and columns anticipated, filled with NA. The assignment of FX_returns[, i] will fill the object all rows at a time for each column and the results will persist.

2. If FX_returns has two dimensions, as indicated by FX_returns[,i], no fewer and no more may be assigned to it. If it is actually a vector it has only one dimension and the assignment should be indicated FX_returns[1]` and no more can be stuffed. So, you should look at each term being assigned to it and make sure they result in a vector.