Abstract
This thesis focuses on a comprehensive analysis of the sources of business cycle fluctuations in the United Kingdom (UK), based on Structural Vector Auto-Regressions (SVARs) with long-run restrictions. We first investigate the sources of historical economic fluctuations in UK labour productivity and employment from 1855 to 2014. To do this, we use a long UK dataset produced by the Bank of England (BoE). Under long-run identification restrictions, we decompose shocks into technology and non-technology shocks, where technology shock is the only permanent shock affecting labour productivity. By identifying structural breaks, we find that the impact of a technology shock on labour is negative, but statistical significance depends on the sub-sample period. We found insignificance for the post-World War II period regardless of labour input. Also, we found that non-technology shock is better at capturing recession dates, and this is consistent with the US. Thus, the technology-driven Real Business Cycle (RBC) hypothesis is rejected for the UK. The analysis also reveals important changes in the volatility of shocks over time, and it is interesting how different sources or events have influenced the economic fluctuations in the UK over time. Secondly, we analyse the effects of technology shocks on outputs and labour inputs across UK industries. For that purpose, we estimate sectoral VARs in which technology shocks are assumed to be responsible for permanent changes in each industry’s productivity. Using three different measures of productivity: i) labour productivity, ii) Total Factor Productivity (TFP) and iii) ‘purified’ TFP, we estimate unconditional and conditional correlations between labour and different measures of productivity. Our results suggest that the effects of technology improvement vary considerably across different UK industries; however,overall the effect of productivity shock on hours is mostly contractionary regardless of the measure of productivity we use at a sectoral level, in line with the aggregate level, as opposed to the situation in the United States (US) as in Chang and Hong (2006). Finally, we further our study on the relationship between productivity and the labour market in the UK since 1970. The main result for the US was that there was a large sign switch of co-movements of correlations of productivity and unemployment in the mid-1980s from negative to positive, close to zero. For the UK we found that there was also a large sign switch, but unlike in the US this moved from positive to negative in the mid-1990s. These changes are reflected in the impulse responses to identified shocks. Thus, to understand why there was a sign switch we use a New-Keynesian’s search model of unemployment as in Barnichon (2010) with nominal rigidities and variable labour effort, by which we compare the effects of technology shocks (i.e. aggregate supply shocks) to the contribution of non-technology shocks (i.e. aggregate demand shocks). We found that, in order to generate impulse responses to fit the empirical responses, we need a more sclerotic labour market for the UK, especially in terms of higher vacancy cost for the later part of the sample, which is inconsistent with the current conditions of the labour market in the UK.