While economists and policy-makers continue to debate the best way to manage the economic reintegration of east and west Germany, there is one point on which most would agree: so far, at least, things have not gone as well as they might.

From the start, the basic problem was that eastern aspirations – for jobs, wages and social provision – could not be realized without massive transfer payments from western taxpayers. Wage convergence between east and west has surged ahead. As a result, German unification has thus far featured a combination of rapid increases in the level of eastern wages relative to eastern productivity, and massive transfer payments from the west. The result has been predictable: enormous unemployment in the east together with rising taxes and a large budget deficit in the west.

Could things have turned out less painfully? Yes, argue Andrew Hughes Hallett, Yue Ma and Jacques Mélitz in a recent Discussion Paper. But first, any discussion of the policy options for the transition must take account of the unavoidable trade-off between the pace of wage convergence and the level of unemployment.

The authors simulate a modified version of the IMF's MULTIMOD econometric model which tracks output, capital stock, employment and unemployment in the eastern länder over the period 1993-2003 (on the assumption that transfer payments are withdrawn by the end of this decade). They can then measure the trade-off between wage convergence and unemployment in the east.

The results are startling. Even if the pace of wage convergence is slowed so that eastern wages are permitted to creep up steadily to a plateau of 90% of western levels by the year 2000, the eastern rate of unemployment in 2003 will still be 16% – more than twice as high as in the West. Simulations in which wages are restrained, so that the eastern unemployment rate falls to the western level within ten years, reveal a steadily growing wage disparity between the eastern and western länder over the next decade (followed by eventual convergence). These results do not depend on any particular pessimism concerning economic performance in the East. On the contrary, the simulations show rapid growth of around 5% a year in the East over the decade and capital accumulation at twice that rate.

The driving force behind the unpleasant trade-off between wage convergence and unemployment is the initial inefficiency in the east. Indeed, the authors estimate that labour-intensive eastern production techniques mean that the nature of the unemployment problem is already disguised by sizeable subsidies to inefficient employment. These subsidies amount to about one-third of aggregate Eastern output, partly in the form of employment subsidies paid through the Treuhand but also through large transfers from capital to labour in the east. The dilemma for policy-makers is painful: if wage catch-up must continue, eastern unemployment will remain high for a very long time, while convergence of unemployment rates means instead that the gap between Eastern and Western wages will increase significantly for a decade.

Do policy makers therefore face an unpleasant but relatively straightforward choice between the 'economic' objective of efficiency and the 'political' objective of convergence? The simulations suggest otherwise: the effects of pursuing one goal rather than the other are not symmetrical. Pursuing economic efficiency first allows fuller convergence. The same degree of initial wage restraint which allows unemployment to converge after a decade still allows 90% wage convergence ten years on. Pursuing wage convergence first would mean not only persistently high unemployment in the east but also substantial burdens to western taxpayers (at least over the period during which they are willing to finance wage convergence). Germany therefore provides a striking example of the adoption of precisely the wrong ordering: convergence ahead of efficiency.

So how can the balance be shifted in a politically acceptable way from wage to unemployment convergence? And what of the case for shifting the bias of western support from capital subsidies and unemployment pay to temporary employment subsidies? Might this not allow incomes to converge faster than do firms' wage costs, thus encouraging firms to hire workers and reducing the unemployment rate, as suggested by Begg and Portes?

Simply increasing subsidies to labour, Hughes Hallett and his colleagues argue, is surely not the answer, given the huge transfers already taking place from capital to labour in the east. Moreover, simply increasing public transfers to eastern labour in return for wage restraint risks impeding labour market flexibility and blunting the incentive to shift to western production methods.

On the other hand, Hughes Hallet, Ma and Mélitz find that simply withdrawing all forms of western support for eastern employment at an early stage (say in 1996 rather than 2003, as assumed in the simulations) has disastrous consequences for output, employment and wages in the east, with unemployment rising above 30% by 1997. The key policy question is therefore not whether western support continues but whether it does so in a way which both slows wage convergence and, by reducing the marginal cost of labour, increases the efficient level of eastern employment. Which is, of course, precisely what the proposed temporary employment subsidies are designed to perform.