Taxation and Labour Supply

C. V. Brown

• Publisher: Taylor & Francis
• Year: 2018
• eBook ISBN: 9780429655852
• Edition: 1
• Topic: Business
• Subtopic: Business General
• Index: Business

First published in 1981. This book reports on a decade of research into the effects of taxation on the supply of labour. In addition to their work in making labour supply estimates, the study explores a number of the ways labour supply estimates can be used. When budget constraints are non-linear it is not possible to estimate the effects of (tax) or other policy changes from knowledge of labour supply elasticities alone, and it is necessary to re-estimate the original model used to derive the estimates. The implications of labour supply estimates for the study of inequality and optimal taxation are considered. Macro-economic models of the economy typically omit labour supply functions or include functions which are inconsistent with micro-economic work on labour supply.

This book will appeal to academic economists, senior students and policy-makers in the field of public finance and labour economics, who will find much of interest from both the theoretical and policy standpoints.

Chapter 1

Introduction

This book reports on more than a decade of SSRC sponsored research on the effects of taxation on the supply of labour. The data is from a national survey in which over 2,000 weekly-paid workers were interviewed. Parts One and Two of the book discuss the data and provide our labour supply estimates. Part Three discusses some of the implications of labour supply estimates.

The importance of labour supply estimates hardly needs stressing: they are obviously important to an understanding of (1) the effects of the tax/ transfer system on the amounts of work done by various groups of individuals, (2) the distribution of income both before and after tax, (3) the size of the work force and the level of unemployment, (4) the level of national income and (5) the level of welfare.¹

ORIGIN OF THE PROJECT

My own interest in the effects of taxation on incentives started accidentally soon after my arrival in Britain in the mid-1960s. Driving home one evening I decided to work out the net income I would receive for some extra-mural lecturing. While I knew that the standard rate of tax was 8s 3d and that earned income relief was two-ninths, the resulting sum was not an easy one to do in my head. It occurred to me that some people might ignore earned income relief and assume that their marginal tax rate was 8s 3d (41.25 per cent) rather than 6s 5d (32.08 per cent). This led to an interest in misconceptions. It was fairly easy to establish that there were widespread misconceptions about tax (Brown, 1968) and this raised the question of whether these misconceptions might affect behaviour.

Work on the project has evolved in a series of stages. The first was in a mini survey on misconceptions which was supported by the Carnegie Foundation and confined to the Glasgow area. Second was a pilot/ feasibility study for the present project, which was supported by the SSRC and was again confined to the Glasgow area. The third (main) stage is the subject of this book and was again supported by the SSRC. For this stage a national survey was conducted in 1971 in which over 2,000 weekly-paid workers were interviewed.

RESEARCH STRATEGY

This section outlines our initial strategy and compares it with what we have done. Research on labour supply has changed dramatically since the strategy for this project was initially formulated in the late 1960s. In the 1960s economectric work on labour supply was in its infancy and many of the problems discussed in subsequent chapters were not known. The initial research strategy involved the use of two largely independent methods of analysis. One, the most common at the time (following Break, 1957), was to start from people’s perceptions of their behaviour and then to examine these statements for plausibility. The other approach that we planned (which has in the event become the dominant one, both in terms of our work and in the work of others) has been to examine behaviour directly using econometric techniques. It was hoped that if the approaches gave consistent results this would increase confidence in the findings.

This decision to use two approaches has had a significant effect on our work. In the interview approach it is important to have ways of checking people’s statements about the effects of taxation on their behaviour. This meant that certain information had to be collected before questions about the effects of taxation could be asked. This in turn meant that (in general) only one interview could be attempted per household. The obvious cost of this approach is that we have inadequate data for household models. While the two approaches have not led to noticeably inconsistent results, the difficulties of comparison were not fully recognized which explains why we have said very little about it.

While in very broad plan this strategy has been followed through, there have been many changes. These have in part resulted from the rapid progress in the subject already referred to. A number of early ideas have not worked out. An attempt to improve the accuracy of our data by checking income data with employers was abandoned (see Chapter 2). Most disappointing to me personally is that we have made no significant progress in the study of misconceptions. On the other hand we have achieved a number of things beyond our original intentions. The detailed econometric developments outlined in Part Two were not envisaged and our initial plans did not include the discussion of the implications of the work that is the subject of Part Three.

PLAN OF THE BOOK

Part One – The Data

One respect in which my own view has not changed is the importance of good data. While there are econometric techniques that can correct for some kinds of data deficiencies these techniques are limited in their effectiveness. A great deal of the effort within the project has gone into collection and handling of data. Despite this effort the data has serious deficiencies and we try to make the reader fully aware of the problems. We believe that one of our most important findings is that estimates (particularly of the elasticity of substitution) are very sensitive to model specification and to data quality. Even though our data falls short of the ideal by a large margin, it is very much better than many of the data sets used for labour supply estimates (many of which were collected for other purposes). In particular our data is sufficiently rich that we have been able to experiment with a number of alternative ways of defining budget constraints. Chapter 2 discusses the sample design and the questionnaire, and Chapter 3 contains a general discussion of data quality. For the benefit of readers who may wish to examine our data base more closely we have included on microfiche copies of the questionnaire, full details of the sample design and copies of the fieldwork documents.

Part Two – The Results

Part Two contains our labour supply estimates. The chapters within this part appear in the order in which the work was done (except for Chapter 9 which was written about the same time as Chapter 6). While the chapters are largely self-contained, Chapters 6 – 9 are to some extent dependent on Chapter 5 and do not repeat many definitional matters discussed there.

We hope that these chapters make a useful contribution to the analysis of the labour supply of British workers. However, all of the findings should be treated as provisional. There are a variety of reasons (discussed in detail below) for this including truncation bias, small cell and incomplete response bias; numbers which are small relative to the techniques employed, lack of adequate household data, and measurement problems – particularly for non-employment income. It is likely to be some years until firmly established labour supply estimates are available for British workers.²

Part Three – Implications

Part Three of the book consists of six almost wholly independent chapters which explore several implications of labour supply elasticities. While some of these chapters make reference to the findings in Part Two, they are not dependent on these findings. Taken together they not only suggest the importance of labour supply estimates, but also the large amount of work that remains to be done in theoretical model building before the implications of labour supply estimates can be fully understood.

NOTES: CHAPTER 1

1  It is perhaps worth reminding those readers who are not professional economists that an increase in labour supply and an increase in welfare are not the same thing – most importantly because welfare includes leisure.

2  Since the autumn of 1979 we have been working on a new labour supply study, financed by HM Treasury, in which the Office of Population Censuses and Surveys will collect on our behalf a larger data set which we hope will overcome the deficiencies of our 1971 data.set.

PART ONE THE DATA

Chapter 2

Sample Selection and Questionnaire Design

This chapter contains summary information about sample selection and fieldwork, together with a discussion of the design of the questionnaire. The Appendix (documents on microfiche) contains a full description of the sample design, of the fieldwork procedures and a copy of the questionnaire. Where possible a distribution of responses is given for each question for both men and women. It is hoped that this will give the interested reader a feel for the data. Preparation for all the matters discussed in this chapter occurred more or less simultaneously but most of our time in the early stages was spent on the development of the questionnaire, which is discussed at the end of the chapter.

It was recognised at the start of the project that different groups of workers such as managers, professional people and self-employed businessmen may respond in rather different ways to changes in taxes to the response of ‘ordinary’ workers. ‘Ordinary’ workers are more likely to be paid by the hour and frequently there is a premium hourly wage if more than a certain number of hours are worked. One of our earliest decisions was that it would be impractical to study all groups of workers simultaneously. We decided to concentrate on workers rather than managers because we thought it would be easier to measure both their wage rates and their labour supply. We also decided to concentrate on people who were at work. We recognized at the time that this decision meant ignoring the participation decision but we did not realize it would potentially cause truncation bias (see Chapter 3).

Having decided to study workers our next problem was to define what we meant by the term ‘worker’. We required a definition that was arbitrary rather than judgmental as the decision about eligibility had to be made quickly on the doorstep by a large number of professional interviewers. We noted that managers tended to be paid monthly and workers weekly. This was a simple, clear criteria which in the event worked well. It avoids the truncation bias which would have occurred if we had used income as our criteria for inclusion. However it excludes workers who happen to be paid fortnightly or monthly.

We also had to decide what we meant by at work. We wanted to include both full-time and part-time workers but wanted to exclude retired and other people who perhaps worked only for the occasional hour. The definition we adopted was people who usually work eight or more hours a week and who had been to work sometime in the seven days prior to interview. This decision leads to truncation bias. To reduce costs we decided not to interview in Northern Ireland or North of the Caledonian Canal in Scotland.

Our sample population is therefore weekly-paid employees in Great Britain South of the Caledonian Canal who usually work at least eight hours a week and who had been employed in the last seven days.

BRITISH MARKET RESEARCH BUREAU

The number of interviews contemplated required the use of professional interviewers. Given that the services of Government Social Survey (as the Office of Population Censuses and Surveys was then called) were not available to us, this meant making use of a market research agency. Extensive enquiries revealed a consensus that there were two agencies which could be expected to reach satisfactory standards on the complex fieldwork required. The two agencies were asked to estimate for the work and, after consultation with the SSRC, the British Market Research Bureau (BMRB) was awarded the contract. BMRB’s contract included sample design, drawing of the sample, production of all necessary documents¹ (based on a fully piloted draft questionnaire that we provided), the fieldwork staff, coding, punching and the production of the data set on cards. There was extensive consultations at all stages of these procedures.

SAMPLE DESIGN

After advice from Professor G. Kalton, the following self-weighting sample design was produced. Two hundred constituencies were sampled. This was the BMRB master sample which was selected with probability proportional to size (PPS) after ranking constituencies within standard regions by the proportion of labour vote. Two areas in each constituency were selected by PPS and twenty-four addresses in areas were selected again by PPS. The names of all electors at each selected address were then listed and two numbers were assigned to that address. The first number, called the interval number, was equal to the number of electors at the address and the second number, the starting number, was a random number between 1 and the interval number.

This information was then used to establish which people should be interviewed. This sift procedure was carried out by the interviewer and any responsible adult at the address. She first lis...

Table of contents

• Cover
• Half Title
• Title Page
• Copyright Page
• Original Title Page
• Original Copyright Page
• Contents
• List of Documents
• Acknowledgements
• 1 Introduction
• Part One the Data
• Part Two the Results
• Part Three Implications
• Bibliography
• Index