README.Rmd

---
output:
  github_document:
    toc: FALSE
---

# Estimating Monetary Sentiment

This package provides a monetary dictionary and appropriate function to measure the latent dimension in central bank communications. The dictionary is a `quanteda` dictionary object with two dimensions, *dovish* and *hawkish*. The innovation of this dictionary is that it contains appropriate valence shifters where they are applicable, thus minimizing false positives. The theoretical underpinnings of the dictionary is in Máté Á, Sebők M, Barczikay T (2021) The effect of central bank communication on sovereign bond yields: The case of Hungary. PLoS ONE 16(2): e0245515. [https://doi.org/10.1371/journal.pone.0245515](https://doi.org/10.1371/journal.pone.0245515)

The `get_monetary_sentiment()` function is designed to work with the dictionary and return a document feature matrix (DFM) which can be used in the downstream analysis.

## Installing

The package is currently only available from GitHub, so it can be installed with the `devtools::install_github()` approach.

```{r eval=FALSE}
devtools::install_github("aakosm/monetary.sentiment")
```


## Usage

The dictionary is included in the package as `dictionary_monetary_sentiment`. The workflow was designed to integrate into `quanteda` as demonstrated below on a small example.


```{r message=FALSE,warning=FALSE, eval=FALSE}
library(quanteda)
library(monetary.sentiment)

# example texts
text <- c(
  "The increasing unemployment is a worrysome development that we have to monitor.", 
  "The current economic climate seems to be stable on the short to medium term.", 
  "The business sentiment is at an all times high.", 
  "The low inflation raises concerns."
)

# create tokens
monetary_toks <- corpus(text) %>% 
  tokens()

get_monetary_sentiment(monetary_toks, dictionary = dictionary_monetary_sentiment)
```