The ESRC Centre at IFS has made many major academic contributions since its inception in 1991. We are at the forefront of bringing rigorous and empirically grounded microeconomic analysis to bear on important academic and policy issues. The importance and wide range of our work are evidenced by the breadth and depth of our publications.
Labour supply, tax and benefit reform, inequality and intra-household allocations
Estimating reliable measures of labour supply responses to tax and benefit changes by individuals and families has been a central aim of research at the Centre since its inception, as they are key components in understanding the impacts of tax and welfare reform. The increased use of means-tested benefits and tax credits to encourage paid work while maintaining redistributive objectives makes it increasingly important to understand the incentives individuals face to gain and retain employment. This is especially true for workers at lower incomes and for individuals in families with children. Our research has developed new estimates of individuals' and families' labour supply responses, which have been used extensively to study the impact of policy reforms [11, 18].
The UK has been a good place to study labour supply effects, due to the major tax reforms in the 1980s and the changing dispersion of wages. Nevertheless, the changing composition of the labour market, macroeconomic shocks, the changing composition of the tax-paying population, and discontinuities in the tax system have created serious estimation problems. Our research developed estimators that help address these issues [11, 13, 1, 10], and have produced precise and robust estimates of the sensitivity of hours worked to changes in wages, which have been used extensively in policy analysis. This research estimated the incentive effects of income taxation in a life-cycle model of consumption and labour supply. The changing dispersion of wages, income, and wealth, also provided an important object of study in itself, with new work characterising and explaining the unprecedented changes in inequality occurring in the UK in the 1980s and 1990s [19, 20].
Traditionally, most empirical work on labour supply decisions has been based on the 'unitary' model - assuming that individuals in families act as a single decision-maker. However, the ways in which individuals within households make joint decisions is a crucial issue. For example, in-work tax credits change the non-earned income received by husband and wife, and thus their financial incentives to work. The collective approach to decision-making has been widespread in the theoretical literature, and research at the Centre has made two important contributions to its empirical implementation. The first was to establish the collective restrictions on labour supply functions and to contrast them with restrictions implied by the usual 'unitary' framework . This helped us to model the facts that we often see non-participation and discrete labour supply in the collective setting. The second contribution was to extend the collective model of household behaviour to allow for the existence of public consumption goods (goods that many individuals in the household consume), which is necessary to understand the allocation of resources to children .
Consumption, savings and pensions over the life cycle
The ageing population is one of the key challenges facing the welfare system of developed economies and in particular their pension arrangements. To address this challenge and the dramatic implications that it would have for the welfare of future retirees, we need to understand how individuals and households make consumption, saving (particularly pension savings) and retirement decisions. Centre researchers have made important methodological and analytical breakthroughs in understanding household consumption dynamics and saving over the life cycle. The approaches that now dominate the field internationally are defined and heavily informed by these insights and techniques.
Our key contribution came in a series of papers that developed and implemented methods for modelling dynamic consumer behaviour using repeated cross-section household-level data, while simultaneously controlling for a large range of other factors. These papers, for example [26, 32, 43], revived interest in the life-cycle model, and showed that an extended version that modelled properly the changing needs over the life cycle implied by changes in family composition, allowed for interactions between labour supply and consumption and uncertainty could explain the main patterns we see in household-level data. This result was an important driver of a host of international academic and policy studies. The life cycle model is behind the idea that the provision of public pension can partially crowd out private savings. Work at the Centre used data on pension reforms in Italy and the UK, respectively, to test this idea and found a substantial amount of crowding out [27, 31]. Subsequent research provided further support for more sophisticated versions of the life-cycle model. We showed that limited participation of households to the stock market can explain the poor empirical performance of asset pricing models based on full participation. This result is explained by the fact that the consumption behaviour of stock market participants is very different from that of non participants .
Our work on the life cycle model, has lead to two further important innovations. The first is a new interpretation of the link between income processes and consumption dynamics. We showed how evidence on the evolution of consumption and income inequality can be used both to identify the role of different types of formal and informal insurance mechanisms in smoothing out various types of shocks and how to separate permanent and transitory components of these shocks [28, 42]. This work now underpins the expanding empirical field that studies the link between consumption and income inequalities. The second innovation used dynamic programming techniques (as well as the preference parameters estimated in our early contributions) to find approximate numerical closed form solutions (where analytical solutions are not possible without making implausible assumptions) for the consumption function allowing for demographic change and income risk . This was one of the first applications of such methods to this problem, which has now become the dominant methodological approach in the field.
Consumer demand, indirect taxes and the cost of living
The Centre has constructed empirical models of consumer demand that are consistent with the observed expenditure patterns of individual consumers over a long time horizon and that also enable us to provide a detailed welfare analysis of shifts in relative prices and indirect taxes. Our research first established that aggregate data alone are unlikely to produce reliable estimates of the structural price and income coefficients that are necessary for welfare analysis . We went on to show that when considering how household expenditure on a set of goods varies with income, it is important to allow this to vary non-linearly . Models that fail to account for this were shown to generate important distortions in the patterns of welfare losses associated with an indirect tax increase . We derived a new and popular model, called 'QUAIDS', which generalised existing models and has since become the standard framework for modelling consumer demand.
Another significant development in our research was to apply the idea that preferences of consumers can be revealed by their purchasing habits (revealed preference theory) to the statistical analysis of consumer demand without imposing parametric assumptions. We showed that by using additional information (about the Engel curve), we could improve the power of non-parametric tests of revealed preference . This can be useful in obtaining precise estimates of, for example, true cost-of-living indices, while making only very few assumptions about household and individual preferences.
A big problem in modelling consumer behaviour is the very large number of decisions faced by consumers. Our research has considered plausible approaches to reduce the dimensions of this problem, including new forms of separability  and models in which consumers value goods based on a limited set of characteristics (characteristics models). Characteristics models have been found to be useful in many areas of economics. However, their empirical implementation tends to rely heavily on functional form assumptions. Our research has developed methods that allows us to relax many of these assumptions .
Training, education and human capital
Human capital accumulation lies at the heart of many explanations of growth, inequality, poverty and crime. Governments spend significant resources on interventions to promote human capital acquisition. The Centre has pioneered research in estimating the returns to education and evaluating the effectiveness of policy interventions in achieving various outcomes. Our research has addressed the returns to education and training for the individual, the firm and the economy at large. It has attempted to estimate the true causal effect of education and training on individual earnings, controlling for potential biases in the estimated returns. It has also explored the question of how the quantification of the individual wage gain from education feeds into policy, and in particular the contribution it can make to the issue of underinvestment in education.
We have used British birth cohort panel data to examine the impact that degree-level qualifications, other higher education and basic educational qualifications have on the earnings of individuals in the medium to longer term [73, 74, 77, 78]. Our research has also examined work-related training and the effect it has on the subsequent employment prospects of men and women. We have examined the wage pay-offs to different types of work-related training, and whether such training improves the wage prospects of relatively low-skilled individuals [72, 79].
Research at the Centre has examined the design and implementation of national programmes in the UK - e.g. the New Deal for Young People (NDYP) , the Education Maintenance Allowance (EMA) [70, 76, 81] and the Excellence in Cities programme  - in other OECD countries - e.g. the Swedish Active Labour Market programmes  - and more recently in developing countries. In our work on the NDYP, we found that the impact of the programme significantly raised transitions to employment by about 5 percentage points. The impact is robust to a wide variety of non-experimental estimators. However, we also presented evidence that this effect may not be as large in the longer run.
This work has opened up a whole new line of research where programme evaluation is considered in the broader context of understanding human capital accumulation and labour market behaviour. We have developed models of education choice, career progression and job mobility, which allow us to understand wage growth and the way that policy can affect the long- and short-run decisions to accumulate human capital through vocational and other education, as well as through on-the-job accumulation of experience .
Microeconometric and evaluation methods
Our work on policy evaluation has led to a number of technical innovations in evaluation and micro-econometric techniques. These have enabled us and other social science researchers to carry out robust and rigorous analysis in a number of areas. We have developed and extended a number of evaluation methods - in particular, a difference-in-differences estimator combined with matching that balances the sample in all comparison groups; this is especially important when each group (before and after, treatment and control) includes data from separate random sampling. In addition, we have developed results to identify effects in models where the response of individuals is heterogeneous and where treatments may themselves be determined by factors affecting the outcome (i.e. may be endogenous) .
Our work on dynamic panel data methods has developed earlier, now standard, estimators  to obtain better performance with panel data that are highly persistent, which is commonly found in economic applications. Estimators that impose mild restrictions on initial conditions can greatly improve precision and reduce finite sample bias . Applications to firm-level production functions confirm these theoretical and experimental results . These methods have been made widely available in software routines and are now incorporated into standard statistical packages like Stata. We have also developed panel data methods for analysing the process by which earnings are determined that separate income shocks into transitory and permanent components .
Wealth creation, firm investment and innovation behaviour
Centre research has made important contributions to our understanding of the impact of corporation tax, financial constraints and uncertainty on firms' incentives to invest, of the determinants and implications of firm location decisions , and of how competition affects firms' behaviour.
Theoretical models of firm behaviour have emphasised the importance of financial market imperfections as a constraint on corporate investment. A series of papers [104, 110, 112, 114] applied new econometric methods to firm-level data to estimate the impact of financial constraints on firm behaviour. More recently, research at the Centre has highlighted the impact of uncertainty in the economic environment as an important influence on firms' investment behaviour , and studied the impact of corporate taxes in the presence of uncertainty. [108, 109]
Research at the Centre has been at the forefront of developing both theory and empirical evidence on the relationship between competition, innovation and growth.  Previously, theoretical work gave ambiguous predictions about the impact of competition on innovation and growth, and had been at odds with empirical work. Research at the Centre has brought both new theory and new empirics to bear on the issues of what impact market liberalisation and other pro-competitive reforms have had on innovative activity and economic performance. [99, 107, 117] Specific contributions of work at the Centre have been: first, to use firm-level panel data to allow us to control for other (often unobservable) factors, and thus empirically distinguish between the different theoretical models; second, a methodological contribution on how to control for heterogeneity at the individual level that is unobserved in models where the observed outcome is a count (for example, of innovations or patents) [93, 106]; third, to develop new theoretical models that reconcile the previous theoretical work with the empirical regularities we observe.
The transfer of technology is also a major driver of growth. Work at the Centre has made important contributions to our understanding of how firms decide where to locate [103, 113], and how information flows across countries and between firms [115, 116]. The Centre has also built up new data sources - matching firm-level accounting data to administrative data from patents offices in the US and Europe - which have allowed us and other researchers to investigate these issues in more depth.
Policy and practice
See all associated publications