Repository logo
 

Nowcasting Inflation in New Zealand

aut.embargoNo
aut.thirdpc.containsNo
dc.contributor.advisorGalimberti, Jaqueson
dc.contributor.advisorVermeulen, Philip
dc.contributor.authorAustin, William Bernard
dc.date.accessioned2023-07-09T21:39:02Z
dc.date.available2023-07-09T21:39:02Z
dc.date.issued2023
dc.description.abstractInflation is a crucial statistic for Kiwis to know. However, Kiwis do not regularly receive official updates on New Zealand's (NZ’s) inflation levels. Statistics on some components of inflation are published monthly. Despite this, statistics on the overall inflation rate in New Zealand are only published quarterly. As a result, it is not easy for Kiwis to know the current inflation levels in New Zealand (NZ). The inflation rate is calculated using the New Zealand Consumer Price Index, referred to as the NZ CPI or CPI. In my research, I look at how best to nowcast inflation in New Zealand. I create model nowcasts from six different models and adjusted them for various situations. I also source professionals and household nowcasts of inflation from nine data sources and compare them with model nowcasts. My research finds that univariate models such as ARIMA (autoregressive integrated moving average) and the RBNZ (Reserve Bank of New Zealand) Target model have similar root mean squared errors (RMSEs) compared to multivariate models. The best multivariate model in my research depends on the month of the nowcast. Specifying the models with a rolling window results in a higher root mean square error (RMSE) value than using a recursive approach. This could be because of the quarterly nature of the NZ CPI, which meant means there is not enough data available to the rolling window models to create a good nowcast. Models that used all the data I had available had access to 16 independent variables. From these independent variables, models are created using only a portion of the data. Specifically, models were made using only soft, economic activity, price, financial, and Phillips curve data. The RMSEs from these models are, on average, similar to or better than when all 16 variables were used. This result may be because adding all the variables into one model may result in parameter proliferation. Models in my sample generally have a lower RMSE value in quarters where the New Zealand inflation rate was between 0% and 1%. This result may be because many models exclusively produce nowcasts within this range. When the published quarterly inflation rate is outside this range, the RMSE value, on average, increases by around 150%. Generally speaking, nowcasts for my six models perform best during quarter 1 and worst in quarter 4. The mean nowcast of inflation by Kiwi households has an RMSE value significantly higher than the best performing univariate benchmark in my sample, the ARIMA model. This finding is consistent with the literature. Meanwhile, the nowcasts of most professionals beat the ARIMA model’s nowcasts. However, the one-quarter-ahead forecasts of four out of the seven professionals have a higher RMSE than the ARIMA model nowcasts. The best nowcasts for the NZ inflation rate have a relative RMSE value of around 60% lower than the ARIMA model nowcasts.
dc.identifier.urihttp://hdl.handle.net/10292/16390
dc.language.isoen
dc.publisherAuckland University of Technology
dc.rights.accessrightsOpenAccess
dc.titleNowcasting Inflation in New Zealand
dc.typeThesis
thesis.degree.grantorAuckland University of Technology
thesis.degree.nameMaster of Business

Files

Original bundle

Now showing 1 - 1 of 1
Loading...
Thumbnail Image
Name:
AustinW.pdf
Size:
1.23 MB
Format:
Adobe Portable Document Format
Description:
Thesis

License bundle

Now showing 1 - 1 of 1
Loading...
Thumbnail Image
Name:
license.txt
Size:
897 B
Format:
Item-specific license agreed upon to submission
Description:

Collections