Essays in empirical macroeconomics and finance

Thesis

This thesis comprises three self-contained chapters. All relate in some way to expectations about macroeconomic or financial variables.

Chapter 1: We study forecasts for real Gross Domestic Product growth by international organisations. These forecasts influence policy decisions at international organisations and national governments, and receive significant attention from the private-sector. We make three contributions. First, we compare the accuracy of forecasters and test the rationality of forecasts using a large dataset of forecasts by four international organisations (the International Monetary Fund, the World Bank, the European Commission and the Organisation for Economic Cooperation and Development). Second, we derive a new decomposition of differences in forecast accuracy into contributions from disagreement and the accuracy of the average forecast. Finally, we present theoretical results and empirical evidence on three explanations for the rejection of rationality tests: the use of conditioning assumptions, the practice of making modal forecasts, and incentives to make biased forecasts for countries receiving assistance from the organisation.

Chapter 2: Forecasts of period-average exchange rates require the use of disaggregated data to both construct efficient forecasts and to correctly test the null of no-predictability. However, in a survey of the literature, we show that all existing studies of periodaverage exchange rates failed to do either. To rectify this, we construct real-time vintages at the daily frequency for all measures of exchange rates and for all countries. We document that the end-of-period no-change forecast, which reflects the null of no-predictability, substantially outperforms the period-average no-change forecast used as a benchmark in existing studies. Moreover, we find that forecasting models are more accurate at short horizons when estimated with end-of-period or daily inputs than with period-average inputs. Finally, we show that models using such high-frequency information almost always outperform the period-average no-change benchmark. In contrast, when these models are compared against the end-of-period no-change benchmark, we find little evidence of predictability at short horizons for nominal bilateral exchange rates, but some evidence of predictability for real and effective exchange rates.

Chapter 3: The theoretical literature has not reached a consensus on whether share prices are affected by taxes on dividends paid to domestic shareholders. This thesis chapter conducts an event study of the announcement of a reform to Australia’s dividend imputation system. The reform would have abolished refunds to taxpayers with excess imputation credits. This would have raised taxes on dividends paid to Australian shareholders with excess imputation credits. The results provide no evidence that the announcement reduced share prices.

Authors: McCarthy, M

• DOI: 10.5287/ora-bp1kpkb2b
• Type of award: DPhil
• Level of award: Doctoral
• Awarding institution: University of Oxford
• Language: English
• Keywords: franking credits; GDP; forecasting; dividend imputation; exchange rates
• Subjects: Time-series analysis; Corporations--Taxation; Forecasting; Macroeconomics; Finance