JOB CREATION AND THE 1999 REFORM OF NATIONAL INSURANCE.

Hart, Robert A. ; Ruffell, Robin J.

Robin J. Ruffell [*]

A radical reform of National Insurance was implemented in April 1999. One objective of the reform was to stimulate job creation. This note analyses the effect of the changes in employers' contributions on employment. We find that the reform could stimulate the creation of about 200,000 jobs in manufacturing industries.

Introduction

In March 1998, Chancellor Gordon Brown announced major changes to the structure of National Insurance Contributions to take effect in April 1999. The Treasury's press release of 17 March 1998 claimed that "[the] radical package of measures will improve incentives, encourage job creation and cut down on red tape." In this note we focus on estimating the jobs created by the changes in employers' contributions.

The structure of employers' contributions

The reform introduces three major changes in employers' contributions:

(a) Contributions are now payable only on earnings above the Lower Earnings Limit (LEL). Under the previous system, contributions were payable on all earnings for any employee earning more than the LEL. This reduces the employer's contribution by [pounds]1.92 per week for all workers earning more than the LEL.

(b) The LEL has been raised from [pounds]64 per week to [pounds]81 per week.

(c) A single contribution rate of 12.2 per cent replaces the stepped schedule (with rates of 3 per cent, then 5 per cent, 7 per cent and finally 10 per cent) that has applied to employers' contributions since 1985.

As has been the case with contributions since the late 1960s, the new proposals differentiate between employees not contracted out and those contracted out of the State Earnings-related Pensions Scheme. Here we describe only the scheme for those not contracted out. The contracted out scheme has essentially the same structure but both the old rates and the new ones are lower for those contracted out than those not contracted out. In line with the Treasury press release, our calculations treat the reform 'as if' it had been introduced in April, 1998, i.e. a year earlier than its actual implementation. In fact, in April 1999, the earnings limits were slightly higher in nominal terms (for example, the LEL is [pounds]83 not [pounds]81).

Comparison of the marginal schedules suggests that the new scheme will have a detrimental effect on jobs, particularly among the low-paid. Thus, for example, the new 12.2 per cent rate replaces a rate of only 3 per cent for those earning between [pounds]81 and [pounds]110 per week. This impression is, however, misleading because the net change in contribution must also take into account the fact that the first [pounds]81 is zero-rated where previously each marginal rate applied to all earnings. So, any evaluation of the reform must take into account the new flat-rate contribution above the LEL and the exemption below the LEL.

A simple and appealing way of expressing these two effects is as follows. Under the new scheme, with Y denoting weekly earnings in pounds, the employer's contribution is 0 if Y [less than] 81 or 0.122(Y - 81) = 0.122 Y - 9.88 if Y [greater than or equal to] 81. Thus, for those employees earning more than [pounds]81 per week (a very high proportion in all industries), the new scheme can be viewed as providing a decrease in their weekly fixed employment costs (f) of [pounds]9.88 combined with an increase in variable costs (v) of 12.2 per cent on all earnings above the LEL. In effect, therefore, the 1999 reform alters the ratio (p = f/v) of fixed-to-variable labour costs. The size of the change in this ratio will vary by industry, depending on the location and dispersion of the earnings distribution. Estimates of the fixed and variable cost changes by industry are shown in Table 1.

Predicted employment effects

In order to estimate the size of the employment changes resulting from the change in p, we use the standard workers-hours labour demand model. By differentiating between the intensive and extensive margins of the firm's operation, this model shows that a change in p has employment consequences. Ignoring output and capital stock effects, a fall in p reduces the marginal cost of employing an extra worker relative to the marginal cost of employing a worker for an extra hour and hence encourages the firm to increase employment.

Hart and Ruffell (1998) presented the results of estimating this model using 1984 data from the Labour Costs Survey. For the present study, we have re-estimated the model using the latest comparable data, relating to 1992. In the interests of brevity, we do not report the full results here. [1] They are generally very similar to the 1984 results. The key parameter estimate needed for the subsequent estimates is the elasticity of employment (N) with respect to the ratio of fixed to variable costs (p). This is -0.28, as compared with -0.35 for 1984.

Before presenting our employment estimates, we should draw attention to some limitations of our approach:

(a) The analysis treats labour as homogeneous. Given that we predict differential impacts on high- and low-paid employees, a complete analysis would require more evidence on factor complementarity and substitutability than is available.

(b) The analysis ignores supply-side influences in general and the supply-side effects of employee contributions changes in particular. Our implicit assumption of an infinitely elastic supply curve will tend to produce over-estimates of employment effects. See Pissarides (1998) for a relevant discussion of the supply-side constraints.

(c) Employers' contribution changes under the 1999 reform are revenue neutral. However, the employees' contributions changes were not and the overall cost of the reform is [pounds]1.4 billion. Our partial equilibrium approach stops short of measuring the opportunity costs of this net expenditure and the associated employment ramifications.

(d) Owing to data constraints, our estimates relate only to the manufacturing, mineral oil processing, water supply, and construction industries.

Table 1 shows the percentage change in employment in each 2-digit industry resulting from the change in employers' contributions, ceteris paribus. These are derived by applying the elasticity of the demand for workers with respect to p to the change in this cost ratio resulting from the change in contributions. The mean fall in fixed costs ([delta] f) over all the industries in Table 1 is 14 per cent. Of course, high fixed-cost industries, such as Mineral oil processing (SIC 14) and Motor vehicles and parts (SIC 35), experience below-average falls in fixed costs while the opposite is true of low fixed cost industries such as Textiles (SIC 43) and Footwear and clothing (SIC 45). Changes in variable costs resulting from the reform are less dramatic, mainly because of the dominance of wages within total variable costs. The all-industry mean change in variable costs ([delta] v) is 2 per cent, ranging from 1.46 per cent in the high-wage Mineral oil processing industry to 3.42 per cent in the low-wage Footwear and clothing industry. All industries experience a rise in variable costs because the new unified rate of 12.2 per cent is above the highest rate under the previous scheme.

The combined effect of the fixed and variable cost variables is such as to produce a fall in p in all industries. Not surprisingly given the universal fall in p, all industries display positive employment effects. For the manufacturing industry, we estimate an employment gain of 203,000 on the 1998 level.

The absence of non-wage data by earnings category means that we are unable to estimate directly the job effects by wage level. We know, however (see Hart, 1984), that high wage employees tend to have higher associated fixed costs (e.g. private fringes and training expenditures). Therefore, the fall in fixed costs due to contribution changes would be expected to have a smaller overall impact on fixed costs for these workers. We can also expect that the rise in variable costs will have a smaller proportional impact on high wage-earners because the rise in the marginal rate is lowest at the top end of the earnings range. Table 1 shows, for example, that in Mineral oil processing, where there is a high mean weekly wage ([pounds]593), there is a 4.9 per cent fall in fixed costs due to the contribution changes and a rise of only 1.5 per cent in variable costs. By contrast, in Footwear and clothing at the other end of the wage spectrum ([pounds]215), the fixed cost fall was 27.8 per cent and variable costs rose by 3 .4 per cent. Thus, the changes in fixed and variable cost both tend to reduce the ratio of fixed to variable costs relatively more at the lower end of the wage distribution. Therefore, we would expect, ceteris paribus, higher job creation among low-paid employees.

Conclusions

In line with Dicks and Robinson (1985), we estimated modest employment increases stemming from the 1985 reform of National Insurance (Hart and Ruffell, 1998). Essentially, new employment was stimulated by this earlier intervention through a combination of removing the Upper Earnings Limit (in all but name) and implementing a steeply progressive rate structure. The downside of this earlier legislation was that it resulted in a complicated structure of rate-bands, and failed to remove the anomaly of marginal rates well in excess of 100 per cent at each rate change.

The 1999 reform is a further positive step forward. First, it simplifies significantly the contribution schedules. One advantage to employees of the new structure compared to the old is that it removes the incentive for employers to avoid wage increases that significantly increase marginal payroll costs due to a jump from one rate-band to the next. Second, from a job creation perspective, the subject of principal concern here, it reduces the fixed costs of employment thereby improving the incentive to increase the employment stock. Like the earlier 1985 intervention, the 1999 reform will almost certainly entail a preponderance of relatively low-paid new jobs. This occurs because induced reductions in the ratio of fixed-to-variable costs are larger among low-wage employees. Unlike the earlier legislation, however, there should be employment creation throughout the wage distribution.

(*.) Comments should be addressed to the authors at Department of Economics, University of Stirling, Stirling FK9 4LA, Scotland.

Note

(1.) A paper presenting the results in full is available from the authors. The model was described in full in our paper on the 1985 reform of National Insurance: Hart and Ruffell (1998).

References

Dicks, G. and Robinson, B. (1985), 'A budget for sterling and a budget for jobs', Economic Outlook 1984-1988, The London Business School, 9, 6.

Hart R. A. (1984), The Economics of Non-wage Labour Costs, London, Allen and Unwin.

Hart R. A. and Ruffell, R.J. (1998), 'Labour costs and employment policy', Notional Institute Economic Review, 165, pp. 99-108.

Pissarides, C. (1998), 'The impact of employment tax cuts on unemployment and wages: the role of unemployment benefits and tax structure', European Economic Review, 42. pp. 155--83.

 Employment effects by industry (ranked by
 % employment effect)
 Industry (SIC8O) Earnings Employers' NI
 [pounds] per week %
 mean sd old new
14 Mineral oil processing 593 295 9.9 10.6
26 Man-made fibres 359 122 9.9 9.4
35 Motor vehicles and parts 409 201 9.9 9.8
25 Chemicals 434 309 9.8 9.9
36 Other transport equipment 403 175 9.9 9.8
17 Water supply 408 194 9.9 9.8
22 Metal manufacturing 391 227 9.8 9.7
21/23 Extraction of minerals 381 182 9.8 9.6
33 Office machinery 470 276 9.8 10.1
32 Mechanical engineering 375 214 9.8 9.6
37 Instrument engineering 338 190 9.6 9.3
24 Non-metallic mineral products 332 171 9.6 9.3
34 Electrical engineering 354 217 9.6 9.4
47 Paper, printing and publishing 390 255 9.7 9.7
41/42 Food, drink and tobacco 333 234 9.5 9.3
31 Metal goods n.e.s. 326 206 9.6 9.2
48 Rubber and plastics 323 181 9.6 9.2
50 Construction 355 209 9.7 9.4
49 Other manufacturing 291 226 9.2 8.9
46 Timber, wooden furniture 295 175 9.4 8.9
43 Textiles 270 161 9.1 8.6
44 Leather, leather goods 274 148 9.2 8.7
45 Footwear and clothing 215 163 8.3 7.7
21-49 all manufacturing 356 227 9.6 9.5
 [delta]f [delta]v [delta]p [delta]N
 (%) (%) (%) (%)
14 -4.9 1.5 -6.3 1.8

26 -6.8 1.5 -8.1 2.4
35 -8.5 1.7 -10.1 3.0
25 -9.8 1.7 -11.3 3.4
36 -10.0 1.7 -11.5 3.5
17 -10.8 1.6 -12.2 3.7
22 -11.5 1.8 -13.1 4.0
21/23 -11.8 1.8 -13.3 4.0
33 -12.8 1.8 -14.4 4.4
32 -13.2 1.9 -14.8 4.5
37 -13.9 2.1 -15.6 4.8
24 -14.7 2.0 -16.4 5.1
34 -14.9 2.0 -16.6 5.2
47 -15.1 2.0 -16.8 5.2
41/42 -15.6 2.1 -17.3 5.4
31 -16.2 2.1 -18.0 5.6
48 -17.1 2.1 -18.8 6.0
50 -17.8 2.0 -19.4 6.2
49 -20.0 2.5 -21.9 7.1
46 -22.3 2.4 -24.1 8.0
43 -22.4 2.6 -24.3 8.1
44 -27.4 2.4 -29.1 10.0
45 -27.8 3.4 -30.2 10.5
21-49 -14.1 2.0 -15.8 4.9
Note:
[delta]f is the change in fixed costs,
[delta]v the change in variable costs,
[delta]p the change in the ratio of fixed to variable costs,
[delta]N the change in employment.