|
on Forecasting |
| By: | Mr. Sakai Ando; Mr. Taehoon Kim |
| Abstract: | Forecasting a macroframework, which consists of many macroeconomic variables and accounting identities, is widely conducted in the policy arena to present an economic narrative and check its consistency. Such forecasting, however, is challenging because forecasters should extend limited information to the entire macroframework in an internally consistent manner. This paper proposes a method to systematically forecast macroframework by integrating (1) conditional forecasting with machine-learning techniques and (2) forecast reconciliation of hierarchical time series. We apply our method to an advanced economy and a tourism-dependent economy using France and Seychelles and show that it can improve the WEO forecast. |
| Keywords: | Macroframework; Conditional Forecasting; Reconciliation; Accounting Identities; Hierarchical Time Series |
| Date: | 2022–06–03 |
| URL: | http://d.repec.org/n?u=RePEc:imf:imfwpa:2022/110&r= |
| By: | Rueben Ellul; Germano Ruisi (Central Bank of Malta) |
| Abstract: | This paper describes a dynamic factor model for the Maltese economy. The model mainly serves as a tool to timely provide the Central Bank of Malta with nowcasts as well as short-term forecasts of the growth rate of the real gross domestic product, which in turn are used as an input in the forecasting process. Such forecasts reflect and incorporate the flow of information that periodically becomes available. Furthermore, the model can handle mixed frequencies that are likely to exist in large datasets used to summarise the Maltese economy and, as an additional advantage, it is able to deal with any path of missing data. This last feature is of crucial importance as data releases that are used to update the model do not take place in a synchronous way. The forecasting power of the dynamic factor model is compared with those of several other models available at the Central Bank of Malta. Overall, the results point towards a higher forecast accuracy of the dynamic factor model at very short horizons while, at longer ones, bayesian vector autoregressions appear to be more reliable. |
| JEL: | C53 E37 |
| Date: | 2022 |
| URL: | http://d.repec.org/n?u=RePEc:mlt:wpaper:0222&r= |
| By: | Waychal, Nachiketas; Laha, Arnab Kumar; Sinha, Ankur |
| Abstract: | In this paper, we propose a new approach for generating a customized forecast ensemble that considers the user's preferences across multiple criteria. The proposed algorithm takes inputs from the user and computes a set of optimal weights assignable to the n chosen criteria. Using these weights, we define a metric called the Multi-Criteria Value, which is maximized to obtain the customized ensemble. This algorithm is called Adaptive Ensemble Generator since it incorporates m distinct forecasting methods and n evaluation criteria. We demonstrate this algorithm customized for four different users (three real-life users and one designed user) on a large database. |
| Date: | 2022–06–30 |
| URL: | http://d.repec.org/n?u=RePEc:iim:iimawp:14678&r= |
| By: | Alexandros E. Milionis (Bank of Greece and University of the Aegean); Nikolaos G. Galanopoulos (University of the Aegean); Peter Hatzopoulos (University of the Aegean); Aliki Sagianou (University of the Aegean) |
| Abstract: | One of the most important risks in the actuarial industry is the longevity risk. The accurate prediction of mortality rates plays a crucial role in the management of the aforementioned risk. Such predictions are performed by modelling the mortality rates using mortality models. Aiming at possible improvements in such forecasts, in this work we examine the effect of data transformation and “linearization†on the quality of time series forecasts of mortality rate data. By the term time series “linearization†is meant the treatment of causes that disrupt the underlying stochastic process measured by a time series. The dataset consists of the time series of the period indices uncovering the mortality trend for England-Wales according to published mortality models. Results indicate a clear improvement in interval forecasts. However, the result on point forecasts is not as clear as is the case of interval forecasts. The documented improvement in interval forecasts can significantly affect the Solvency Capital Requirement, and subsequently the Solvency Ratio for a pension fund. Such an improvement might put some pension providers at a competitive advantage as they have less capital locked in their liabilities. In addition, it was confirmed that the transformed-linearized time series of mortality rates satisfy to a higher extent the need for normality as compared to the original series. |
| Keywords: | Time series transformation and ‘’linearization’’; Outliers; Actuarial time series forecasts; Mortality rates; Covid-19 |
| JEL: | C22 C51 C53 C87 G22 |
| Date: | 2022–05 |
| URL: | http://d.repec.org/n?u=RePEc:bog:wpaper:297&r= |
| By: | Robert Ambrisko |
| Abstract: | Macroeconomic data are published with a time lag, making room for nowcasting macroeconomic variables using fiscal data. This is because a) monthly and daily fiscal data are available from the state budget in a very timely manner and b) many fiscal data are the function of macroeconomic variables. I employ two nowcasting models, bridge equations and MIDAS regressions, which link quarterly macroeconomic variables to monthly fiscal data for the Czech Republic. Bridge equations are found to be particularly suitable for nowcasting the wage bill using social contributions, achieving a 2% improvement in the root mean square error (RMSE) of one-quarter recursive forecasts compared to historical CNB forecasts. Further, I propose a tractable method for incorporating daily data into the nowcasting models, relying on STL decomposition by Cleveland et al. (1990). Depending on the timing, the RMSE for the wage bill can be up to 4% lower when the available daily data on social contributions are taken into account in the nowcasting models too. |
| Keywords: | Bridge equations, daily data, fiscal, midas, nowcasting, real-time data, short-term forecasting, STL |
| JEL: | C53 C82 E37 |
| Date: | 2022–06 |
| URL: | http://d.repec.org/n?u=RePEc:cnb:wpaper:2022/5&r= |
| By: | Yanzhao Zou; Dorien Herremans |
| Abstract: | Bitcoin, with its ever-growing popularity, has demonstrated extreme price volatility since its origin. This volatility, together with its decentralised nature, make Bitcoin highly subjective to speculative trading as compared to more traditional assets. In this paper, we propose a multimodal model for predicting extreme price fluctuations. This model takes as input a variety of correlated assets, technical indicators, as well as Twitter content. In an in-depth study, we explore whether social media discussions from the general public on Bitcoin have predictive power for extreme price movements. A dataset of 5,000 tweets per day containing the keyword `Bitcoin' was collected from 2015 to 2021. This dataset, called PreBit, is made available online. In our hybrid model, we use sentence-level FinBERT embeddings, pretrained on financial lexicons, so as to capture the full contents of the tweets and feed it to the model in an understandable way. By combining these embeddings with a Convolutional Neural Network, we built a predictive model for significant market movements. The final multimodal ensemble model includes this NLP model together with a model based on candlestick data, technical indicators and correlated asset prices. In an ablation study, we explore the contribution of the individual modalities. Finally, we propose and backtest a trading strategy based on the predictions of our models with varying prediction threshold and show that it can used to build a profitable trading strategy with a reduced risk over a `hold' or moving average strategy. |
| Date: | 2022–05 |
| URL: | http://d.repec.org/n?u=RePEc:arx:papers:2206.00648&r= |
| By: | Sara B. Heller; Benjamin Jakubowski; Zubin Jelveh; Max Kapustin |
| Abstract: | This paper shows that shootings are predictable enough to be preventable. Using arrest and victimization records for almost 644,000 people from the Chicago Police Department, we train a machine learning model to predict the risk of being shot in the next 18 months. We address central concerns about police data and algorithmic bias by predicting shooting victimization rather than arrest, which we show accurately captures risk differences across demographic groups despite bias in the predictors. Out-of-sample accuracy is strikingly high: of the 500 people with the highest predicted risk, 13 percent are shot within 18 months, a rate 130 times higher than the average Chicagoan. Although Black male victims more often have enough police contact to generate predictions, those predictions are not, on average, inflated; the demographic composition of predicted and actual shooting victims is almost identical. There are legal, ethical, and practical barriers to using these predictions to target law enforcement. But using them to target social services could have enormous preventive benefits: predictive accuracy among the top 500 people justifies spending up to $123,500 per person for an intervention that could cut their risk of being shot in half. |
| JEL: | C53 H75 I14 K42 |
| Date: | 2022–06 |
| URL: | http://d.repec.org/n?u=RePEc:nbr:nberwo:30170&r= |
| By: | Michael Kitchener; Nandini Anantharama; Simon D. Angus; Paul A. Raschky |
| Abstract: | This paper proposes a new method to predict individual political ideology from digital footprints on one of the world's largest online discussion forum. We compiled a unique data set from the online discussion forum reddit that contains information on the political ideology of around 91,000 users as well as records of their comment frequency and the comments' text corpus in over 190,000 different subforums of interest. Applying a set of statistical learning approaches, we show that information about activity in non-political discussion forums alone, can very accurately predict a user's political ideology. Depending on the model, we are able to predict the economic dimension of ideology with an accuracy of up to 90.63% and the social dimension with and accuracy of up to 82.02%. In comparison, using the textual features from actual comments does not improve predictive accuracy. Our paper highlights the importance of revealed digital behaviour to complement stated preferences from digital communication when analysing human preferences and behaviour using online data. |
| Date: | 2022–06 |
| URL: | http://d.repec.org/n?u=RePEc:arx:papers:2206.00397&r= |