Errors in survey and administrative data on employment earnings: Austria and the United Kingdom compared (with Christopher R. Bollinger, Stephen P. Jenkins and Fernando Rios-Avila)
We contribute new cross-national evidence about the nature of measurement errors in employment earnings, fitting the same error components model to harmonised earnings data for Austria and the UK. The model allows for measurement error in the administrative data and linkage error as well as survey measurement error. We find several cross-national similarities in error structure but also intriguing differences in error component probabilities, means, and dispersions.
Automatic stabilization: the missing welfare dimension in Latin America (with Olivier Bargain and H. Xavier Jara)
Comparing the redistributive effects of tax-benefit systems across countries can be useful to benchmark national policy design. However, this type of analysis often forget the role of automatic stabilization, i.e. the ability of systems to mitigate income losses in times of downturn. We provide a unique international assessment using tax-benefit simulations associated with household surveys for 50 countries of three large regions (Europe, Latin America and Sub-Saharan Africa). Using a well-established methodology, we show that tax-benefit systems in Latin America outperform those in Sub-Saharan Africa in terms of income redistribution and poverty reduction, but fiscal systems in both regions provide a limited degree of automatic stabilization against income shocks, in absolute terms and relative to Europe. This limited capacity is due to three factors: (i) the prevalence of a large informal sector, which limits the role of social insurance contributions and personal income taxation; (ii) the presence of high tax exemption thresholds and generous tax deductions; and (iii) the fact that cash transfer programs are not means-tested (i.e. mostly rely on proxy means-tests), which prevents them from acting as stabilizers.
The Persistence of Income Reporting Errors in Household Survey Data (with Christopher R. Bollinger)
We examine the extent and persistence of income misreporting in household survey data. We use a unique panel of survey data – the Austrian version of the European Union Statistics on Income and Living Conditions (SILC) for 2008-2011 – which have been linked to individual administrative records on both state unemployment benefits and earnings. We find that misreporting earnings and benefits receipt in one year increases the probability to continue doing so in subsequent years. This causes large errors in transitions in and out of earnings and benefits receipt, leading to a biased picture of earnings dynamics and government programme participation. The persistent error pattern has important implications for policy and research, as we demonstrate by examining poverty dynamics. Since transitions in and out of benefits and earnings receipt are misreported, the survey income reports bias downward estimates for poverty persistence and entry, and upward estimates for poverty exit.
The Effectiveness of Social Protection in Five African Countries through Normal Times and Times of Crisis (with Katrin Gasior and Gemma Wright)
We study the effectiveness of social protection benefits in reducing income and consumption poverty in five Sub-Saharan African countries – Ghana, Mozambique, Tanzania, Uganda and Zambia – in normal times and times of wide-spread economic crisis. Using tax-benefit microsimulation models with representative household survey data, first we estimate the coverage of benefits and their poverty-reducing effects in each country. Second, we study the ability of benefits to stabilise incomes and consumption in times of crisis by simulating hypothetical reductions to earnings and employment. Although the coverage of benefits is fairly high in Ghana and Zambia, the poverty-reducing impact of benefits in all five countries is low in normal times. The effectiveness of benefits to stabilise income and consumption in times of crisis is also limited because many benefits are linked to proxies of income, not income itself, or have tight eligibility criteria. Social assistance programmes are typically unresponsive to losses in household earnings and employment and provide limited support for unemployed people.
Child Poverty in Smaller Families: Why So Little Change and What Would Make a Difference (with Kitty Stewart, work in progress)