Enterprise restructuring and social benefits

Economics of Transition, Volume 5 ( I ) , 1-24, 1997

Enterprise restructuring and social benefits Simon Commander and Mark Schankerman'

ED1 World Bank 1818 H Street Washington D.C. TEL:202 473 6293

European Bank One Exchange Square

London EC2A 2EH

TEL:0171 338 6904

Abstract

Soviet era firms provided generous social benefits, including health and child care. Despite recent cuts, firm survey data show that benefits have remained a major component of total compensation. With benefits largely finn-specific and firms dominated by insiders, continuing attachment of workers as well as widespread informal sector participation has resulted. This has impeded restructuring, in part by generating significant set-up costs for new private firms. We simulate the effects of a cut in subsidies to benefits provision. We show that while this leads to falls in benefits and employment and an increase in wages, the outcome critically depends on the availability of alternative providers. The key to cushioning these adverse consequences is the stimulation of a market in benefits provision. Given initial conditions, rapid removal of benefits supports will require transitional income support to avoid under-consumption of these goods. We provide the design of a simple scheme of transitional support and show that it can be financed from the savings from removal of current subsidies to benefits.

JEL Classification: 523, 532, J38, P31. Keywords: social benefits, restructuring, subsidies, enterprise behaviour.

1. Introduction

Firrrs were the primary locus for providing social protection in the Soviet system. The legacy is widespread ownership of social assets, the main components of which are housing, health clinics and kindergartens. In general, users of firm-supplied services have received these services gratis or for very nominal fees and as a consequence cost recovery rates have remained very low. The corollary of this has been low cash wages. In effect, the previous system attempted to balance benefits provision by a lower cash wage, while raising attachment through firm-specific benefits.

This structure of compensation has continued into the transition with significant implications for the speed of restructuring and allocation of effort by workers. In this paper we argue that the linkage between employment in state firms and the provision of

2 Enterprise restructuring and social benefits

social benefits, or other non-monetary components of compensation, creates attachment of workers, retards labour mobility and slows restructuring. The current subsidies to enterprise-provided social benefits, reflected in low cost recovery rates, induce insider- controlled firms to adjust on three margins - wages, employment and benefits. The mix of adjustments depends critically on the availability of outside, competitive providers of benefits. We formally model and simulate these adjustments, and argue that the impact of eliminating benefit subsidies to firms will depend on how quickly competitive provision of benefits develops. Moreover, given the dominant control by insiders in state and privatized firms in many transition countries, including Russia and Ukraine, this will tend to promote the growth of an informal private sector as workers allocate effort between their original employers and private activity. We argue that while this may indeed smooth employment adjustment in a state sector suffering significant shocks, it will impede the pace of restructuring and is likely to have adverse fiscal implications.

The paper discusses the appropriate roles of firms, households and government in the supply of and demand for social benefik2 Although private firms in OECD countries often provide significant non-monetary components of compensation, indicating some room for discretion in terms of the locus of supply, it is generally accepted that former Soviet firms provided far higher shares of benefits in compensation than either their OECD or eastern European counterparts. This has resulted in firm boundaries that are ill- suited to a market environment, and likely created efficiency costs in the provision of social benefits, since efficiency in production may be very different from efficiency in benefits supply. One obvious approach to this issue has been to promote privatization of firms along with divestiture of social assets.3 But shifting part of the burden from firms may simply have displaced the locus of the financing and management problem. One manifestation is that housing subsidies in Russia exceeded four per cent of GDP in 1995, on a rising trend.

This paper extends the discussion of enterprise provision of social benefits in several respects. First, it pulls together the empirical evidence on provision and the costs of provision, drawing primarily on two surveys of Russian and Ukrainian firms. Second, the paper takes a closer analytical look at the issues and their implications for public policy. We analyse the effects of social benefits provision by firms on worker attachment and restructuring and the associated questions regarding efficiency in both the supply of and demand for benefits. We argue that keeping provision in existing firms and restricting access to insiders retards restructuring and distorts relative efficiency, in part by generating significant set-up costs for de novo firms who need to offer such benefits in order to attract and retain workers. In short, the paper argues that the current system sponsors attachment, retards restructuring and strongly skews the private sector toward informalization. Not least of all, it allows firms to retain an implicit financing claim on government.

An effective reform must reconfigure access to benefits by de-linking provision from incumbency in a firm. This offers the potential for generating a market in provision and partially addresses the problem of outsiders, workers who have lost their jobs. It limits the moral argument for benevolence, by no longer making access to social protection contingent on employment, and consequently facilitates restructuring. De-linking employment and access to benefits does not require that firms divest social assets to local municipalities or some other government authority.

For merit goods - principally health and child care - there are grounds for believing that firms will not necessarily choose to maintain benefits once subsidies are withdrawn, particularly if firms are constrained in pricing benefits (as is often the case now). But there is also the basic problem of the inability to adjust instantaneously the

Commander and Schankerman 3

structure and level of compensation. This will generate under-consumption of such merit goods through the channel of income inadequacy. This suggests the provision of transitional income support. Such income support would decline over time as wages adjust, but some form of means-tested support could remain in place to address the needs of poorer consumers. We show that transitional income support can be provided with no net drain on the government revenues, relative to the existing subsidies for social benefits, and discuss alternative mechanisms for providing that support, in particular consumer vouchers and producer rebates. Concluding remarks summarize our findings and policy recommendations.

2. Benefit provision in Russia and Ukraine: survey evidence

The results presented in this section are primarily drawn from two enterprise surveys. The first was a representative survey carried out in mid-1994 in Russia covering 435 industrial firms.4 The second was a smaller, non-random survey covering 59 mainly industrial firms in Ukraine that was implemented in September 1995. Both surveys capture a mix of privatized and state firms. However, from the viewpoint of this paper, perhaps the most important feature regarding ownership has been the continuing dominance of insiders in decision-making and control, in privatized as well as state firms.' One important implication is that there is, to date, little identifiable difference in behaviour between ownership categories.

Prior to presenting the main findings from these surveys, it is worth emphasizing several common features in the evolutions of' these two countries since the break-up of the Soviet Union. First, official figures indicate cumulative declines in GDP and industrial output of the order of 50 per cent since 1990 in both countries. These substantially overestimate the decline in GDP by not capturing the informal private sector. Nevertheless, controlling for changes to output, those to employment have been quite limited. The picture that emerges from both contexts is of limited involuntary separations with, however, quite significant adjustment to hours through either short time or involuntary leave. Real wages by 1995 had fallen by over 50 per cenf from pre- transition levels, although there are notorious measurement issues here. Consequently, unit labour costs in both Russia and Ukraine have declined substantially. Broadly, it appears that workers have, wherever possible, traded down the monetary component of compensation for employment stability.

Part of the sluggish adjustment in employment can be attributed to continuing transfers to the enterprise sector. Budgetary subsidies in 1995 accounted for between 3-6 per cent of GDP for Russia and Ukraine respectively - a clear decline over 1992/93 - but these arc gross under-estimates as they exclude directed credits, tax arrears and other sources of soft finance from both local government and the banking system. At the same time, general government revenue relative to GDP fell in both cases and most sharply in Russia where the decline has been from 42 per cent in 1992 to 27 per cent in 1995. While part of this reflects lower levels of activity, it also results from systematic tax evasion on the part of enterprises and households. Indeed, a major factor behind this evolution has been the growth of an informal private sector largely outside of the tax net. The negative fiscal implications of the gradualistic adjustment to restructuring have consequently been substantial.

Further, budgetary supports to the provision of social budgets have continued, even when, as in Russia, over 50 per cent of the housing stock had been privatized by mid-


1995. At that time, subsidies to local utilities for the supply of energy, water and waste disposal alone amounted to 6 per cent of GDP, of which over 4 per cent was from local governments. This points to a continuing feature in both countries; low cost recovery ratios for social benefits. For instance, in Russia, housing costs in 1995 accounted for no more than 2-3 per cent of average household income.‘ Cost recovery for other services again seems to be low, with very limited change over time. This can primarily be traced to the absence of incentives for firms to raise recovery rates, including the presence of subsidies for benefits provision, as well as the imposition of caps on cost recovery by government.

Table 1. Major benefits: scope of provision, 1990/91 and 1994/95 (% of firms)

Benefit Type mid- 1994 1990/9 1 mid- 1995 1991

Russia Ukraine

Childcare/childcare subsidy 66 79 45 59

Healthcare 70 71 95 91

Housing or housing subsidy 55 59 47 47

Construction of new housing 50 73 50 65

Table I summarizes the survey evidence on the prevalence of benefits provision by firms in Russia and Ukraine, both for 1990/91 and 1994/95. We concentrate on the major benefits - childcare, healthcare and housing - but firms have also commonly provided holiday homes, subsidized food and other minor benefits. Table I shows that a large majority of firms provided the major benefits before transition and most have continued to do so. But there is evidence that the provision of childcare and new housing construction has declined both in Russia and Ukraine. Further, in both countries a significant share of enterprise housing - in excess of 25 per cent - had been privatized by 1994. But the countries differ in the extent to which the scale of provision has been curtailed over the same period (for details see Commander, Lee and Tolstopiatenko, 1996). By 1994/95 between 25-35 per cent of firms had maintained a comparable scale of benefits relative to pre-transition in both countries, and in Ukraine the majority of firms had either increased or maintained provision. However, there is clear evidence that a significant share of firms in Russia have begun to adjust the scale of benefits. Nearly 40 per cent of Russian firms providing benefits in 1994 had cut provision very substantially, but there was no systematic association between reductions and ownership. And while roughly 20 and 30 per cent of firms in Russia and Ukraine, respectively, reported benefits provision as a ‘major financial burden’, nearly 50 per cent in each sample cited them as a necessary ‘social burden.’ At the same time, firms claim that ensuring worker attachment was a relatively minor factor influencing decisions on the scale of benefits. However, it is unclear whether enterprise managers distinguish sharply between considerations of worker attachment and ‘social responsibility.’

Table 2 shows that in Russia, in mid-1994, state-owned firms and privatized firms were very similar in terms of the scope of benefit provision. Between 50 and 75 per cent of such firms provided each major benefit. While a much smaller fraction of de nova firms offered these benefits, more than 25 per cent of them provided childcare and healthcare. To examine the characteristics of benefits-providing firms more formally, we ran a set of descriptive, ordered probit regressions of the number of benefits (up to eight) against a set of firm characteristics. The exercise is conducted for each country using data for 1991 and 1994/95.


Table 2. Major benefit provision and ownership, mid-1994 (% enterprises) Benefit type State-owned Privatized De novo

Childcare 68 69 23

Healthcare 75 72 27

Housing 56 58 9

New housing construction 53 51 11

Turning first to Russia, the leading finding is that larger firms are more likely to offer benefits. The estimated coefficient on the size of the firm, measured by (logarithm of) employment, is positive and highly significant, and it is not sensitive to the inclusion of dummy variables for profitability, ownership, and location. Moreover, the point estimate is very similar in the 1991 and 199415 data. The second finding is that the number of benefits is also positively related to the level of monetary compensation (average wage). The point estimates of the wage coefficients are again similar for the two periods (for the same specification), but they are only statistically significant for 199415. If firms were trading off the level of monetary compensation for benefits, one might expect a negative sign on the wage. But these ordered probits relate to the scope, not scale, of benefit provision. In cross-section data, one might expect that the wage variable may be capturing the effects of the profitability of the firm, through rent-sharing. But this interpretation is not supported by the evidence since inclusion of a dummy variable for profitability (equal to one if there are positive accounting profits) raises, rather than lowers, the wage coefficient. Third, the coefficients on the ownership dummy variables (omitted for brevity), which capture the status of the firm in the later period relative to state firms, are generally insignificant. The only statistically significant ownership coefficient is for de novo firms, which indicates that they are more likely to offer fewer benefits than either state-owned or privatized enterprises, holding size and monetary compensation constant. The raw data indicated that de novo firms are less likely to offer benefits than state-owned or privatized firms (see Table 2). The results in Table 3 confirm that this is at least in part an ownership effect, and is not purely due to the fact that de novo firms are typically much smaller than state-owned or privatized firms. But while they offer few benefits, recall that about 25 per cent of de novo firms in the sample did provide child and health care. This is consistent with evidence from Poland and suggests that such firms need to offer at least some benefits in order to compete for worker^.^ Fourth, firms are more likely to offer a larger number of benefits if they are profitable, but the estimated coefficient on the profitability dummy is not significant in either year. The coefficients on location and dominant industry dummies (not reported for brevity) are statistically insignificant, except for a strong negative coefficient for the major urban areas. Thus, firms in major cities offer fewer benefits, conditioning on firm size and monetary compensation, but there are no other discernible differences due to location or to whether the firm is dominant in the local labour market (e.g., a ‘one company town’). Taken as a group, one can reject the null hypothesis that the ownership, location and dominant industry dummies do not matter. The Chi-square test of regressions I versus 2, and 4 versus 5 confirm this finding. But the rejection is due exclusively to the effects of de novo firms and major urban areas, as noted above.

As in Russia, we find that the scope of benefit provision by firms in the Ukraine is positively related to firm size (see Panel B). The coefficients on the employment variable are about twice as large as for the Russian sample, and they are estimated with sufficient precision in the 1994/95 data that one can reject the null hypothesis that the coefficients for Russia and Ukraine are the same, despite the rather small sample size. But there is


no evidence that benefit provision is related to monetary compensation. The legal form of the enterprise does not significantly affect benefit provision in Ukraine, conditioning on firm size (recall that de nova firms are not available in these data). The Chi-square test on the legal form dummy variables confirms this fact.

Table 3. Scope of benefit provision: ordered probits for Russia and Ukraine, 1990191 and 1994/95

Panel A: Russia 1990/91 1994/95

Variable (1) (2) (3) (4) (5) (6) Log employment .#3* .#O* .439* .397* .427* .427*

(.056) (.052) (.052) (.044) (.040) (.04) Log wage .221 .158 .I51 .259* .I43 .156* *

(.17) (.16) (.16) (.lo) (.097) (.096) Profitability .158 ,134 .2 14 .I51

Ownership yes no yes no

Location and dominant Yes no Yes no N 246 246 32 1 321 Pseudo-RZ .098 .075 .lo6 .085 Test dummies xz (10)/10 2.28 2.8

dummy (.22) (.22) (S9) (.19)

dummies

Panel B: Ukraine 1990/91 1994/95

Variable (1) (2) (3) (4)

(29) (.25) C19) (.18) Log wage .055 .I18 .252 .200

(-82) (.81) (.28) (26) Legal form dummies Yes no Yes no N 30 30 52 52 Pseudo-R2 .I09 .083 .142 .122 Test dummies x2 (3)/3 1.01 1.40

* Statistically significant at the 0.05 level. ** Statistically significant at the 0.10 level.

Log employment 1 .00 .756 .94 1 ,857

For Russia, ownership categories include state-owned, privatized worker-owned, manager-owned, and dominant outsider owned, de novo, and ‘other’. Location categories include major city, oblast centre, rural and ‘other’. The dominant firm dummy variable refers to firms classified as the primary employer in the location. The profitability dummy is one if the enterprise reports positive profits (376 out of 439 firms). For Ukraine, enterprise legal forms include state-owned, company joint-stock enterprise, other joint stock, and ‘other’.

The ordered probit coefficients in Table 3 can be used to estimate the impact of firm size and wage on the probability that any specified number of benefits will be provided. The point estimates imply that the size of the firm has a non-trivial impact on the scope of benefit provision, particularly in Russia. For example, the point estimates for Russia


imply that a one standard deviation in log employment raises the probability that at least two benefits are provided (nearly always child and health care) by about two percentage points in 1991 and 6.6 points in 1995. The corresponding figures for Ukraine are 3.6 percentage points in 1991 but only 0.5 points in 1995.'

Table 4. Change in the scope of benefit provision 1991-199415: ordered probits for Russia and Ukraine

Russia Ukraine Variable (1) (2) (3) (4) Log employment .129* .097** -1.51* - 1 .oo*

(.058) (.053) ( 5 3 ) (.4 1 A Log employment .235* .225* 4.52* 1.34

(.I 13) (.117) (1.93) (1.33)

( . I 14) (.109) ( . 59 ( S O ) Profitability dummy Yes no n/a nla Ownership dummies Yes no n/a nla Location and dominant Yes no nla nla firm dummies Legal form dummies nla nta Yes no N 232 232 28 28 Pseudo-R2 .039 .021 ,402 .224 Test all dummies 0.88 3.27 x2(dYd (d=10) (d=3)

A Log wage .206** .198** .038 -.565

* Statistically significant at the 0.05 level. ** Statistically significant at the 0.10 level. Variables are defined in notes to Table 3. The ordered probits are conducted on three categories of benefit changes: reduced, maintained and increased. A positive coefficient shows that the covariate is associated with lower probability of benefit reduction.

Table 4 reports the results from ordered probit regressions relating the qualitative change in the number of benefits between 1990191 and 199415 to a set of firm characteristics, including the initial employment level, and the change in employment and wages over the period. We group firms into three categories according to whether they reduced, maintained or increased the scope of benefits (the breakdown of the Russia sample is 36, 47 and 17 per cent, respectively; in the Ukraine sample, 16, 60 and 24 per cent). The estimated coefficients for Russia show clearly that larger firms were less likely to reduce the number of benefits, and that firms which raised wages or employment (or reduced them less) were also more likely to maintain or increase benefits. This second finding - that adjustment on wages, employment and the number of benefits tend to move in the same direction - suggests that adjustment was at least partly driven by cash flow, The coefficients on the ownership, profitability, location, and dominant firm dummy variables were jointly (and individually) insignificant. Thus we use the estimates from column (2) to compute impact coefficients. A one standard deviation increase in initial employment is associated with a 5.7 percentage point fall in the probability the number of benefits is reduced. A one standard deviation increase in employment adjustment implies a 5.1 point rise in the probability of benefit contraction; for a one standard deviation increase in wage adjustment the figure is 4.9 points. In short, the initial size of


the firm is important to benefit contraction and Russian firms appear to be adjusting benefits, wages and employment in the same direction during this period.

For these regressions the Ukraine sample is very small (28 firms) due to missing observations, and one should be cautious in drawing conclusions. With this caveat in mind, we note first that the Chi-square test indicates that the firm's legal form does affect benefit adjustment. As expected, the point estimates on the legal form dummies (not shown) indicate that state firms in the sample are less likely to reduce the number of benefits than other, mostly joint-stock, companies. The surprising result is that the coefficient on initial employment is negative and statistically significant. This suggests that larger firms in the Ukraine sample are more likely to reduce the number of benefits, which contrasts sharply with Russia, and probably reflects the relatively soft budget constraints in Ukraine during this period. But the positive and significant coefficient on Alog employment shows that these firms, as in Russia, tend to adjust employment and the number of benefits in the same direction. There is no evidence that the change in benefit provision is linked to wage adjustment over this period. Yet despite their statistical significance, the employment variables are not very important determinants of adjustments to benefits in the Ukraine sample. For example, a one standard deviation increase in employment implies less than a one percentage point increase in the probability the firm reduced benefits, compared to a 5.7 point decline in Russia. The impact of employment adjustment on benefit contraction is similarly small.

In response to shocks, we can think of firms having the option of adjusting along several margins - benefits, wages and employment. The evidence presented above points to very significant stability in the scale and scope of benefits provision. In both samples, firms have adjusted real wages and employment substantially. By mid- 1994 average real consumption wages were down by between 2040 per cent in Russia and Ukraine. In the Ukraine sample, mesn employment fell by around 25 per cent between 1991-94, while in Russia the decline was just under 18 per cenf.'

What have these adjustment choices implied for the structure of compensation? First, in both samples wage and benefit levels were highly correlated both in aggregate and across branches of industry, suggesting that benefits have not generally been used as a substitute for monetary compensation, Relating changes in benefits to those in wages, we find no evidence that firms have systematically used benefits - despite some apparent tax incentives - to offset changes in the monetary component of compensation.

Table 5. Shares of benefits in compensation by ownership, Russia and Ukraine, 1994/95

Russia Ukraine 1994 1994 1995

Ownership Mean Median Mean Median Mean Median State 12.2 9.1 25.9 13.2 24.9 13.7 Privatized 17.5 10.7 26.0 17.6 20.2 13.5 de novo 10.7 7.0 All 15.7 9.1 25.9 15.3 22.3 13.5

Table 5 summarizes information on the share of benefits in total worker compensation, broken down by ownership category (mean values shown are weighted by total compensation). In the Russia sample, benefits on average comprise about 16 per cent of total compensation in 1994. In Ukraine the share is closer to 25 per cent, rising to around 40 per cent in the case of the larger, more benefits-intensive, firms (not


shown). The mean share is much larger than the median in each country, reflecting the skewness in the benefit shares even within ownership categories. There is evidence of a decline in the share of benefits in compensation in the first half of 1995 relative to 1994, the average share falling from 25.9 to 22.3 per cent (also evident in the medians). This decline is concentrated mainly in privatized firms, where the mean benefit share has fallen sharply from 26 to 20 per cent in only one year. The differences across ownership groups in benefit shares are striking. The average benefit share in total compensation is larger for privatized than for state and de now firms in Russia. And this, despite the fact that low wages and wage arrears probably tend to bias the measured benefit share upward for state-owned firms.

Table 6. Share of benefits in Compensation: least squares regressions for Russia and Ukraine, 1994195

Russia Ukraine Variable (1) (2) (3) (4) (5) Log employment 2.90* 2.28* 2.39* 4.60* 4.37*

(.41) ( .36) (.37) ( I .65) ( 1 SO) Log wage .52 .85 1.13 2.14 2.37

(1.12) (1.13) ( 1 . 1 1 ) (2.46) (2.28) Profitability dummy 4.09* 3.43* no nla nla

(1.34) (1.27) Ownership dummies Yes no no nla n/a Location and dominant Yes no no nla nla firm dummies Legal form dummies nla nla nla Yes no N 248 248 248 52 52 Adjusted R2 .I49 .138 .132 ,048 .097 x 2 Test on dummies 0.43 1.63 0.83 x2(d)ld (d=10) (d=l) (d=3)

* See notes to Table 3.

Statistically significant at 0.05 level.

Table 6 reports the results of a simple regression relating the share of benefits in compensation to the employment and wage levels and a set of other firm characteristics. As we would expect, in both Russia and Ukraine the benefits share is positively and significantly related to the size of the firm. And the effect is large: a one standard deviation increase in the log employment level is associated with a 4.9 percentage point rise in the benefits share in Russia, and 4.4 percentage points in Ukraine (using the unrestricted specifications (1) and (4) in the table). There is a positive association with the wage level in both places, but not a very robust link. In Russia the benefits share in compensation is also positively related to whether the firm is profitable, and the estimated effect is large: the benefits share for profitable firms is on average 4 percentage points higher. The ownership variables are generally insignificant, The larger urban areas and those with a dominant firm have lower benefits shares relative to smaller towns, but the coefficients are not significant. The Chi-square tests in the table confirm that the ownership, location and dominant firm dummies are not (jointly) significant determinants of observed variation in the benefits share.


While these results are informative, it is worth emphasizing that the numbers on the share of benefits in compensation are problematic for several reasons. The first is inflation-induced measurement bias. Of course, both the operational cost of benefits (mainly labour) and wages will be understated if there are wage payment lags, but the historical cost of facilities used to provide benefits will seriously understate their opportunity cost, and thus the benefit share (especially for state-owned firms). Even worse, the level of benefit provision is generally measured strictly in terms of the firm's operational costs of providing them, excluding all capital costs of using facilities owned by the firm. Some indication of the extent of this distortion can be obtained by comparing the reported costs of benefits provided by firms in different ownership categories, available in the Russia survey. The reported benefit costs of de now firms were double those of privatized firms and four times the level reported by state firms."' As these large differences are unlikely to reflect variations in the relative efficiency of providing benefits, we interpret them as being, at least in part, an indicator of the unrecorded capital costs associated with benefit provision in state-owned firms. In any case, the figures show clearly the significant costs imposed on de novo firms that provide such services.

Table 7. Cost recovery rates in Ukraine, 1995 (%)

Benefit type Mean Median Number of firms Childcarc 13.9 10.0 21 Healthcare I .8 0.0 44 New housing construction 28.4 5.0 15 Workers' housing 38.7 31.5 20 Holiday homes 28.4 10.0 31 Cafeteria 43.3 32.0 39 Food shop X3.2 I00 21 Transportation 36.1 0.0 15

Aggregate data point to continuing large subsidies to benefits, particularly through energy pricing. The Ukraine survey allows us to get some sense of the distribution of subsidies within the firm by providing more detailed information on cost recovery rates at the level of the firm." Although these rates do not impute market values to the costs of provision, they give some indication of the pricing response by firms. Table 7 summarizes the evidence for different types of benefits. The relevant sample size is small (see last column) and there is significant variation across firms, but on the whole cost recovery rates on the major social benefits remain low, with an average recovery rate of less than 25 per cent in the full sample. However, cost recovery rates are significantly higher for items such as food and cafeterias. And there is some evidence of increased charges for housing (not shown), although these comparisons will be particularly affected by undervaluation on the cost side. Ownership has little effect on cost recovery rates in Ukraine: state firms actually report slightly higher recovery rates than either closed or open joint-stock companies.

Given that cost recovery will depend partly on the ability of consumers to pay and thus on their monetary compensation, we also ran a regression of the cost recovery rate for each type of benefit against average wage in the firm, a set of ownership dummies, and the level of employment.'* We found a consistently positive association between recovery rates and average wage, but it was not statistically significant except in the case of housing. Ownership generally did not matter.


To summarize, the small changes in benefits provision that we observe in Russia and Ukraine can primarily be traced to a combination of continuing subsidies and the absence of alternative institutions (including competitive markets) for providing social protection. While there is some evidence of selective increases in cost recovery, these have remained constrained by the presence of price controls and by the level of cash wages. From the perspective of firms, these choices seem difficult to explain. For firms facing adverse demand shocks and seeking to lower labour costs, we might expect work-sharing, rather than layoffs, to be selected in cases where labour costs varied with hours of work. Conversely, if a component of labour cost, such as benefits, is invariant to hours a firm would have an incentive to reduce the workforce rather than adjust hours. Given the evidence we have presented, why have firms in Russia and Ukraine not responded to a large fixed cost component in compensation by imposing layoffs? Adjustment costs - for example, severance rules - do not appear to be a major explanatory factor. The control structure of firms and the lack of outside mechanisms for social protection seem more promising explanations, which we investigate below. From the perspective of workers, continued access to firm-specific benefits increases the incentive to remain attached. With a lack of outside provision, the marginal value of benefits to workers will tend to exceed the cost to firms. Any decline in the share of monetary in total compensation would reduce the return to individual-specific effort. With incentives to separate from the state or privatized sector being small, a predictable outcome has been the rapid expansion in informal employment and in the overall informalization of these economies.13

3. Attachment and access to benefits

The survey evidence identifies several important issues. The first relates to the dynamics of restructuring. It is evident that with benefits provision remaining relatively stable and tied to state and privatized firms, the incentive for workers to stay in those firms will be enhanced. And by giving existing social benefit providers subsidies and/or tax breaks that compensate for provision, firms will have weaker incentives to reconfigure their benefits and wage structures.

Second, de novo firms will have to provide equivalent compensation packages to these firms to compete in the labour market. This imposes significant start-up costs on private firms and hence tends to restrain the emergence of an autonomous private sector. Insofar as the supply of benefits in the short run is largely restricted to the universe of existing providers, say because of the combination of financing and entry barriers, workers will exhibit little willingness to accept substitution of monetary compensation for benefits. The incentive for workers to remain attached to original providers will be increased. This low substitution from the perspective of workers reflects the absence of alternative mechanisms for benefit provision, and would be expected to increase as alternative sources develop. We return to this important issue more formally in the next section.

A third cause for concern is that restricting access to social services to membership in state or privatized firms creates a clear incentive for private firms to continue to externalize this cost by relying on part-time or moonlight labour. The rapid and pervasive informalization of the economies in Russia and Ukraine is, at least in part, a reflection of this choice. This is likely to have clear adverse productivity effects and, of course, potentially disastrous fiscal implications by lowering yet further the effective private sector tax yield.14


The channels from the structure of compensation to worker attachment, and the implications for informalization of the private sector, can be thought of in the following way. Consider the question of labour allocation in a world where state (or privatized) firms provide benefits, while de now firms can choose whether or not to provide such benefits. Workers in the state sector can work exclusively in the state sector or allocate time across both sectors, so multiple job holding is feasible. The only condition for workers to retain attachment to the state sector, and thus ensure continued access to benefits, is that they work a minimum level of hours. One can show that if the returns to labour in the state sector are sufficiently small, then all who can find a secondary job in the private sector will do so and a corner solution will r e~u l t . ' ~ The de ttuvu firm's choice over whether to hire full-time workers and provide benefits or to rely on part-time workers with access to benefits in the state sector, will depend on a combination of factors including the value of benefits provided by state firms, the substitution elasticity between monetary compensation and benefits, and the set-up costs that such firms would incur in providing equivalent benefits. In short, there will be an incentive for de now firms to shift the cost of benefits provision to state firms, assuming that the unit labour costs (relative to productivity) of part-time workers are not too large relative to full-time workers. The general result is that the current arrangements for benefits provision motivate attachment and informalization in the private sector. Moreover, this is especially likely when there is substantial excess employment labour in the state sector (so returns to labour are small), and when there are limited outside alternative sources of benefit provision (so the worker's willingness to trade benefits for monetary compensation is limited).

This analysis points to several important avenues to lower attachment and shift the incentives for private sector job creation from informal to formal activity. The first step must be to limit the current method of subsidy to benefits providers, given that provision of benefits, at least by the larger providers, appears in general to be met with compensating finance. Apart from the ad hoc nature of these subsidies and tax breaks, they provide the wrong incentives for raising both efficiency in supply and demand for services.

The second issue concerns the need for changing access rules. These are already changing for housing, where between 25-35 per cent of the housing stock is now privatized and where, at least in Russia, much of the remaining housing has been divested to municipalities. But for other benefits, provision is still predominantly tailored to those employed in the firm, despite recent changes in the direction of more open access. Consequently, loss of employment - an obvious component of restructuring - could lead to exclusion. At the same time, by having such exclusion, de novo firms will face start-up costs if they are to compete in the labour market. This distorts competition and is clearly undesirable. For the merit goods with which we are concerned, changing access rules - eliminating current exclusion - is key. It provides the basis for a market in competitive service provision and it changes the incentives for the allocation of private sector jobs between the formal and informal sectors.

4. Benefit provision by incumbent firms

From this discussion it is clear that policies on ownership and control of social assets, primarily those generating merit goods, should be consistent with restructuring. This requires ensuring open access to separate provision from employment status. By dropping exclusion in a multi-firm setting, there are clear opportunities for priming a competitive

Commander and Schankeman 13

market in benefit provision, at least for some types of benefits. Complementary reforms, including the legal separation of service providers from the production units of the original firms, would be required. But several obvious questions arise. The first concerns the dynamic incentive effects that might be expected in settings where there is a monopoly provider, such as a company town, or where there is a small number of providers with a similar branch profile. The second, which is increasingly relevant for both countries, is whether privatized firms have the correct incentives to retain or dispose of social assets, and the implications for efficient provision. We deal with each in turn.

4.2. Company or industry towns Consider the case where employment, and thus service provision, is highly concentrated. In the extreme case of the company town with full employment, there will be an equivalence of interest between the firm and the local community and the level of provision would be the same whether provided directly by the firm or by local government. This equivalence collapses with a departure from full employment and a potential divergence .of interest arises between remaining insiders and those made redundant.I6 Ensuring unrestricted (but not free) access to the social services, reinforced by regulation to prevent discriminatory pricing at the expense of outsiders, addresses part of the problem. But there remains the potential for the dominant incumbent firm to engage in uncompetitive practices, such as cross-subsidizing benefits to restrain the development of a competitive market in provision, particularly for services with significant sunk entry costs. In principle, this might be prevented by requiring the enterprise to keep separate accounting records for the production and benefits ‘divisions’, but this is not likely to be effective given the state of accounting procedures and the limited monitoring capacity of (local) government. In this context, there may be a case for forcing divestiture as a means of ensuring open access.

Further, if benefits provision depends on the viability of the firm or of a small number of firms, this raises serious financing issues. For services with high alternative use values for facilities and a declining share of insiders in total potential users, there will be strong incentives for a firm subjected to a hard budget constraint to cut back on provision. These factors suggest that forcing divestiture in these cases - in the first instance to municipalities - will be important, even if the resolution of the financing issue must be taken up in the overall reform of local government finance.”

4.2. Insider-controlled firms and incentives for provision Turning now to a multi-firm environment - the more common setting - a frequent objection to allowing firms to decide on whether or not to provide social benefits is that left to their own devices they will tend to under-supply such benefits. Clearly, if we believe this assumption to be correct, we may need to consider measures that ensure that provision does not fall too much.

The key question is how firms will respond to a reduction in current subsidies to benefit provision. To examine this issue formally, we conduct a simple simulation exercise which models the behaviour of an insider-controlled firm, recognizing that the firm can adjust along three margins -benefits, wages and employment. We think of the insider-dominated firm as a reasonable characterization of the control structure of the bulk of Russian and Ukrainian enterprises.” The firm maximizes a utility function that depends on employment, wages and benefits, subject to a hard budget constraint. The budget constraint limits total compensation per worker (wages plus benefits, net of subsidy) to the average product of labour. In a pure worker-dominated firm, the level of employment enters the optimization problem only through its effect on the average


product in the constraint. Employment would be set to maximize the average product of labour, say at n*. But we think of a firm starting with an initial employment level higher than nu, so that lowering the employment level toward n' increases the probability that any given worker will be made redundant (homogeneous labour is assumed). Thus, if the utility of remaining in the firm exceeds the outside option, the employment level will enter the utility function directly as well.

There are two basic trade-offs in the model. First, in order to increase total compensation, the average product must be higher and thus employment must be lowered. The second is the trade-off between wages and benefits. The optimal mix will depend on the subsidy rate to benefits and on the elasticity of substitution between benefits and wages in the utility function. The substitution elasticity is a utility parameter, but in this set-up we interpret this parameter as a reflection of the availability of alternative, external suppliers of benefits." If there is no outside market or other source for benefits (such as local government), the willingness of workers to trade off benefits for wages will be small. As outside sources develop, the worker's elasticity of substitution would be expected to rise. By varying the substitution parameter in the simulations, we can examine how the firm's adjustment to a reduction in subsidy depends on the development of outside benefit sources.

Formally, we assume that insiders have a CES utility function in wages and benefits:

U(w, b) = (a wP+ (1-a) bP )"P

where w and b denote wages and benefits per worker and cr =l/( 1-p) is the elasticity of substitution. The probability a worker remains employed in the firm depends on the extent of labour-shedding. If the adjustment from the initial level of employment n,, to the final level n occurs in one period, as is formally the case in this set-up, then the probability that a representative worker remains in the firm should be n/n,,. However, we also experiment with a concave function (n/n,,)* and interpret h< 1 as a simple way of allowing for the employment contraction to take more than one period. Of course, a more complete analysis would model the multi-period decision explicitly. Thus we write the expected utility for a representative worker as:

V(n, w, b) = (n/n,Jh (U(w, b) - x ) + x (2)

where x is the value of the outside option. It is clear that the maximization of V is independent of the value of 4,.

We assume that the firm has a translog production function:

log y = a l o g n + p/2 [log nI2, where pl 0. (3)

This gives an elasticity of average product of (a-l)+p log n, which has the desirable property that it rises (in absolute value) as the degree of excess employment increases - i.e., cutbacks in employment raise average product more sharply when n is large, as one would expect. For ease of interpretation we normalize n'=l, so n is interpreted as the percentage excess employment over the value that maximizes average product. This normalization imposes the restriction a= 1 .*" The firm's budget constraint requires that the average product of labour equal total compensation (net of subsidy):

exp( p/2( log * ) = w + b(1-s) (4)

where the left-hand-side is the average product for the translog and s is the benefit subsidy rate.

Commander and Schankerman

Figure 1. Wages

15

M

d

1.10

0.80 1 0 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.45 0.50 0.55 0.60 0.65 0.70

Subsidy a = 0.9; a = 1; p = -2

Figure 2. Employment

I

c 2 1.00

$ 0.95 w

+ 0 = 0.5 + 0 = l t- a = 2

Subsidy

a = 0.9; 0: = 1; p = -2


Figure 3. Benefits

1.6

1 A

1.2

1 .o

+ 0 = 0.5

3 a

H Oa 0.6

0.4

0.2

0 0 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.45 0.50 0.55 0.60 0.65 0.70

Subsidy a = 0.9; a = 1; p = -2

Figure 4. Benefitdremuneration

0.40 1 + (I = 0.5 0.35

0.30 5

6 0.20

9 8 0.10

'd

5 025

a \

0.15

0.05

0 0 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.45 0.50 0.55 0.60 0.65 0.70

Subsidy

a = 0.9; a = 1; p = -2


The simulations compute the values of employment, benefits and wages that maximize (2) subject to (4), given the set of utility and technology parameters (a, h, (T, p) . Experimentation showed that the qualitative features of the simulations are not sensitive to a, h and p in a reasonable range, but the value of (T is critical. We present results for fixed parameter values a=0.9 and p= -2.00, but allow (T to take on low, intermediate and high values, 0.5, 1 and 2.” Figures 1-4 present the optimized values of wages, benefits, employment, and the share of benefits in total compensation as a function of the subsidy rate, s, for the value k1.0. We also experimented with k0 .75 , but the simulation results were very similar. We want to trace the movement of each of these variables as the benefit subsidy is reduced, while at the same time recognizing that the development of alternative benefit sources will raise the relevant value of (T over time. To make this comparison easier, we normalize all values by the solution corresponding to o=OS and s=SO (we think of these parameters as a reasonable description of the prevailing situation, with limited outside sources for benefits and an initial fifty per cent benefit subsidy). This fixes the reference point at 0=0.5, s=.50, so all solution values are expressed relative to it. This allows us to compare across the curves for different values of (T and identify both the impacts of changes in the subsidy rate and the elasticity of substitution. In particular, as the benefit subsidy is reduced we move down along the low (T curve, but then as outside benefit suppliers develop we jump to a higher (T curve at the then prevailing subsidy rate.

The simulations show clearly that reduction of benefits subsidy will predictably be associated with a decrease in benefits, as well as an increase in wages and a fall in employment. But the key parameter governing the size of those adjustments is the substitution elasticity, (T. When it is small, as one would expect in the early phases of adjustment before the development of alternative benefit suppliers, the composition of compensation is not much affected by the reduction in subsidy. The main impact is to reduce the levels of both wages and benefits, and of employment. But for higher values of 0, reducing the subsidy shifts compensation toward wages and mitigates the negative impact on employment. Thus, as alternative mechanisms to provide benefits develop, the impact of subsidy reduction is to decrease firm-provided benefits and to increase wages, which mitigates the initial decline in employment. In short, the simulations clearly show that the key to cushioning the negative wage and employment effects of reduced benefit subsidy is to accelerate the development of outside mechanisms for benefit provision.

5. Benefit price adjustment and transitional income support

With merit goods we know that one common problem is that even when consumers are fully aware of the benefits of consuming these goods, they may be unable to consume the minimum socially desirable level without unacceptable cuts in the consumption of other basic commodities. Are there grounds for believing that the wage adjustment in response to reductions in benefit subsidy will only be gradual and that, as a consequence, income inadequacy remains a major problem in Russia and Ukraine?

There are two separate issues here, permanent inadequacy for segments of the population and transitional shortfalls as wages adjust. As to the first, household income and expenditure data for both Russia and Ukraine reveal a widespread incidence of poverty and deprivation. Estimates for 1994 give poverty headcounts of the order of 3 2 4 0 per cent for Russia and Ukraine, respectively.22 While we may expect these incidences to change over time, it is evident that significant shares of the population will


require ongoing income support. This involves standard issues of targeting and associated costs. Milanovic’s (1995) analysis suggests that for the transition economies, income testing rather than minimum income guarantees would be a preferable screening mechanism. The second component, which primarily arises from adjustment lags, is more immediately relevant for our analysis.

Adjustments to prices can originate through two channels. First, as firms directly subsidize consumption of social benefits, they will increase their cost recovery rates on the services they deliver as the subsidies and tax breaks for benefit provision decline. Second, both firms and households receive large implicit subsidies through the pricing of utilities. Reducing or eliminating subsidies to energy is essential if the right signals are to be sent to raise efficiency in demand. Given the current level of energy prices in both Russia and Ukraine, any convergence to border prices implies huge adjustments. For instance, complete elimination of the subsidy for gas would raise household tariffs by at least twenty-five times. Projected energy price increases will shift the cost of operating the housing stock alone by around 50 per cent, equivalent to 2.5 per cent of GDP.2’ The size of these adjustments has led to announcements of staggered adjustments. In the housing sector, for example, full cost recovery is projected to be phased in over five years, subject to the constraint that expenditure on housing should not exceed 2 0 p e r cent of household income at the end point.24 The scale of the adjustment and uncertainty over its possible incidence is one factor behind the slow privatization.

The alternative approach that we propose - which is consistent with accelerated privatization in housing - is to remove the benefit subsidies to firms, and from firms to consumers, more rapidly and to accompany this by transitional income support. Because cost recovery rates for most enterprise-provided benefits are very low, rapidly removing these subsidies for benefit provision will substantially raise the relative prices of benefits to workers. If firms are faced with hard budget constraints, this will occur quickly. However, the adjustment to nominal wages will depend on the willingness of workers to trade off wages for benefits (as the simulations illustrate), and thus on the availability of outside sources for benefits, and will only occur gradually. Further, the income effect of the relative price change will depend on the prevailing share of benefits in total compensation and on cost recovery rates. Since the ratio of benefits to compensation and cost recovery rates vary across firms, lagged wage adjustment will induce some reallocation of labour across firms (provided non-exclusivity in benefit provision is enforced).

The income effect reduces benefit consumption in the short run until wages adjust. It will take time for a substantial supply response through expansion and new entry. Unless prevailing levels of benefit consumption are judged to be excessive, some transitional income support will be necessary to prevent an undesirable short-run reduction. The adjoining box describes how to compute the required transitional income support, at least to a first approximation. The method is not informationally demanding and can be easily adapted as the price and benefit levels on which it is based change over time. The proposed measure is an upper bound to the required income support for two reasons. First, it does not take into account any savings that accrue through more efficient competitive provision of benefits, and, second, it does not capture the gains from more efficient consumption of utilities induced by cost-based prices. The box also shows that a policy which fully removes the enterprise subsidy for benefit provision, and provides compensatory transitional income support to households, yields a net fiscal gain. The saving is increased if the government also eliminates the indirect subsidy to enterprises by adjusting utility prices to cost levels and at the same time provides income support to households to neutralize the income effect of the price adjustments. These price


Box 1. Computing the transitional income support

The cost per worker of benefit j is cj , including imputed capital costs. The firm's benefit subsidy rate is sj , so total benefits subsidy per worker is Cs, ci. The government compensates the firm for its benefit provision at rate 8 (in practice, the subsidy is not benefit-specific), so 8=C wj sj where wj=cj/ Xcj. There is also indirect subsidy to the extent utilities are underpriced to firms. If the direct subsidy is eliminated, profits fall by An= 8 X c , ~ . The firm in turn imposes full cost recovery, so real income per worker falls by Am=& c,. Of course, Aa=Am: profits remain unchanged after cost recovery (at old levels of wages, employment and benefits). We assume that the firm bears the burden of the reduced government subsidy, which seems reasonable given the elastic labour supply under present conditions.

The wage adjustment needed to neutralize this income effect is Aw=Am. Suppose that wages adjust slowly over T years, with fraction a, of the required adjustment occurring in period t, Aw,=a,Am. Then transitional income support per worker in year t is S,=( I-%) Am, Under this scheme, each year the government saves the enterprise subsidy 8 C ci =Am and pays out the transitional income support ( l a , )Am. Thus the net fiscal gain is Amza, d, where d is the discount factor. Since this gain depends on how fast wages adjust, government has an incentive to facilitate adjustment.

Budgetary costs of utility price adjustments Cost recovery must occur rapidly to provide appropriate dynamic incentives. This increases the required transitional income support, but it can be fully financed from the reduction in current utility subsidies. And to the extent that demand for utility services is price responsive, there is a net fiscal gain from this scheme, as shown in the figure. The current price po involves a subsidy to the utility of poabp,. Under cost-based prices, the required income support is p,cdp,. The net saving in subsidy cost is abcd. Given current low cost recovery rates, the savings can be large even with very low values of the income- compensated demand elasticity. For example, with demand elasticity of -0.25, moving from the typical 20 per cent cost recovery rate to cost-based prices would reduce the level of consumption (and total subsidy) by one-half. Savings are proportional to the demand elasticity. And rapid adjustment of prices to cost recovery levels provides the right incentives for investments that may be required to curtail consumption (e.g., metering equipment, insulation, efc.).

smc

4


changes in turn provide the proper incentives for competitive provision of benefits (where feasible) and for rational demand-side adjustments.

6. Transitional income support: implementation issues

There are two alternative means of implementing the income support, vouchers for consumers and rebates to benefit providers. For both vouchers and rebates, there is also the choice of the form they take: fees for specified services (with or without price limits), capitation compensation (a fixed amount per person to finance some defined set of services), and fees per casehisit. In making this choice, the key considerations are how they affect supply incentives, demand incentives, and the implementation costs. It is particularly important to mitigate the problem of moral hazard, in the form of inflated costs by benefit providers and excessive use of services by consumers.

Vouchers are relatively easy to administer and preserve customer choice among competing benefit suppliers, which is a key requirement to ensure efficient provision. In addition, they create fewer moral hazard problems on the supply side than rebates because consumer choice will impose some cost discipline on providers. This should be true whether the vouchers or rebates are issued on a fee for service or a capitation basis. However, this discipline will be only be effective in cases where there are multiple sources of supply, or entry is easy - e.g., childcare and small-scale health clinics. But vouchers on a capitation or case basis do not provide incentives to economize on the use of services. Moreover, vouchers have one key disadvantage. They are tradable and thus may be diverted to finance consumption of non-merit goods. It is very difficult to prevent trading effectively, and attempts to do so create black markets which heavily discount vouchers and lead to undesirable wealth transfers. It may also be difficult to specify adequately the range of services for which the voucher is valid, particularly for health facilities, and attempts to do so may discourage supply of innovative services. Vouchers may thus be more easily implemented on a capitation, rather than a fee for service, basis. But capitation vouchers shift the risk of fluctuating demand to the consumer. This may be especially problematic in the case of health provision.

The alternative to vouchers is for the government to provide direct rebates to the benefit suppliers. Rebates on a fee for service basis create very unattractive incentives on the supply and demand sides, leading to cost inefficiency in provision and excessive use of services. The incentives for cost minimization may be marginally improved with price controls, but these may be difficult to specify fully when there are diverse services. They may also encourage quality degradation, especially where there are few competing providers, and they can be skirted by various means including ‘unbundling’ of services. Thus, fee for service rebates may be particularly unsuitable for large-scale health services. Rebates on a case basis provide incentives for cost minimization but not for economizing on use of services. In addition, they may induce deterioration in the quality of provision and encourage providers to select only low risk cases (to the extent they are identifiable ex ante) or else to refer cases. The administrative costs of this scheme are also likely to be high. Rebates on a capitation basis provide good cost incentives but poor ones on the demand side, and they also may encourage providers to degrade quality and take on only low risk customers. In addition, under capitation rebates the benefit provider bears the risk of uncertain or fluctuating demand for services, if mandated treatment is enforced. But capitation rebates have the advantage that they can be relatively easy to administer.

We summarize these considerations in Table 8. This brief analysis suggests that the choice between vouchers and rebates is not clear-cut and may depend on the type of


benefit in question. It is important to emphasize, however, that for any ongoing targeted support scheme for low-income households, rebates have the serious disadvantage that the firm bears the responsibility for ensuring eligibility of the recipient (and hence the risk of making an error). For rebates to work here, the government would need to issue secure certificates of eligibility (with identification to prevent tradability). It is clear that there are different trade-offs involved in each income support scheme. Mixed schemes involving the use of case and capitation would allow for the support to be better tailored to the service being provided. The final design will need to be tailored to country circumstances.

Table 8. Comparison of implementation schemes Vouchers Rebates

Supply Demand Admin. Supply Demand Admin. incentives incentives costs incentives incentives costs

Fee for services + + Case basis + +

basis Capitation + + + +

7. Concluding remarks

The debate surrounding social benefits provision has tended to set up a false dichotomy. There are those who argue for immediate divestiture, while others point to enterprise provision of non-wage benefits in OECD countries as indicating that it would be best to leave it to firms to decide. Neither approach is adequate to address the incentive issues in both the supply of and demand for benefits. We have shown that the current system of provision is inefficient and impedes enterprise restructuring. It imposes significant costs on the economy and is likely a significant factor behind the informalization of the new private sector that has occurred. The fiscal consequences are not pleasant and include more transfers to firms as well as massive tax evasion by the private sector. To break this cycle requires the introduction of an appropriate incentive regime on both demand and supply sides.

Access to benefits is key. Making access open to individuals outside the firm can prime a potential market in services and snaps the link between employment and benefits. It reduces the chance of exclusion as firms restructure and shed labour and thus promotes the labour and capital reallocation that is central to the transition. To enforce open access, even if firms retain control over social assets, they should make benefits divisions independent in an accounting sense. But we are not prescriptive about the locus of provision. Where enforcing access is problematic - as in company towns - it is probably necessary to mandate divestiture in order to achieve open access. Where (mainly large) firms have facilities with large sunk expenditures, such as hospitals, this should be accompanied by mandated divestiture to municipalities.

This reform must be accompanied by rapid reduction of benefit subsidies to firms and removal of remaining price controls on benefits. We simulate the response of insider- controlled firms to benefit subsidy reduction. The simulations show that the decline in benefit subsidy initially induces a decline in both benefits and employment but an increase in wages. As alternative sources of benefits provision develop, the main burden of subsidy reduction falls on firm-provided benefits, while wages rise more sharply and the initial reduction in employment is mitigated. The central policy message is that the


key to cushioning the negative wage and employment effects of reduced benefit subsidy is to accelerate the development of outside mechanisms for benefit provision.

To mitigate the risk of significant under-consumption of merit goods arising from the lagged wage adjustment after removing benefit subsidies, and to make the adjustments politically palatable, it is essential to implement a transitional system of income support. The costs of transitional income support can be fully financed by the savings from removing benefit price subsidies, with net savings to the extent benefit consumption contracts. The income support can be achieved either through vouchers to consumers or rebates to providers. The precise mechanism and associated trade-offs will depend on the nature of the service being provided. The transitional income support programme can be expected to facilitate reforms and to provide budgetary relief relative to the prevailing system.

Endnotes

1 .

2.

3.

4.

5.

6. 7. 8.

9. 10.

1 1 .

The authors are from the World Bank, and the London School of Economics and the EBRD, respectively. We have benefited from comments by the editors on an earlier version of the paper. We thank Geoff Shuetrim for research assistance on the simulations and Sumana Dhar on the empirical work. For earlier work in this area, see Jackman (1995), Schaffer (1995), Boycko and Schleifer (1994), Commander and Jackman (1993) and Estrin et al. (1995). For example, Boycko and Schleifer (1 994) recommend wholesale divestiture of social assets to local government. A full account of the sampling procedure and coverage is given in the paper by Lee in Commander, Fan and Schaffer (1996). The detailed analysis of the Russian data in various papers in Commander, Fan and Schaffer (1996) shows that there is no clear evidence that the behaviour of state and privatized firms differ significantly. See also EBRD, Transition Report (1995), Chapter 6. See World Bank, Country Economic Memorandum (1995a). For evidence on Poland, see Estrin et al. (1995). We compute the impact of a one standard deviation change in a covariate, xk (log employment or wage), on the probability of at least two benefits being provided as Pk f(X’ p) ok, where Pk is the ordered probit coefficient on covariate xt, (T, is the sample standard deviation of the covariate, and the standard normal density function F(.) is evaluated using the estimated threshold value for at least two benefits and the mean values of covariates in the regression. The results are very similar if median values are used. We also computed the implied impact of a one standard deviation change in the log wage on the probability of providing at least two benefits. For Russia the estimate is only 0.2 percentage points in 1991 and 0.9 points in 1995. We do not compute the wage impact for Ukraine since the regression coefficient on wage was not even close to statistical significance. See Commander, Dhar and Yemtsov (1996). The reported mean benefit costs (in roubles per month) are 16,593 for state, 31,682 for privatized and 66,795 for de now firms. See Commander, Lee and Tolstopiatenko (1996). The World Bank Russia survey we use does not contain any information on cost recovery rates. However, a new survey in Russia conducted by the Institute for


12. 13.

14.

15.

16. 17.

18.

19.

20.

21.

22. 23. 24.

Advanced Studies in Vienna provides detailed data on benefit provision and cost recovery. For a description of the survey see Tratch, Rein and Worgotter (1996). The regression results are available on request. An attempt to quantify the size of the ‘unofficial’ economy can be found in Kaufmann and Kaliberda (1995). For a summary of available estimates for various transition countries, see EBRD, Transition Report 1995, Appendix to Chapter 2. We already observe in Russia a strong decline on the revenue side, with falling profit tax and social security collections, together with large-scale tax avoidance by private firms. We have formalized this argument in a two sector model, which is available on request. These issues are explored in more detail in Commander and Jackman (1993). See Le Hoeurou (1995) for a discussion of inter-governmental finance issues. Boycko, Schleifer and Vishny (1995) propose the wholesale divestiture of all social assets to local authorities, not limited to cases of towns dominated by one provider of social benefits. Their approach raises some difficulties. Besides politicizing the provision of these services, local authorities may not be the most efficient providers and, in any case, are not presently in a position to finance them. On this latter point, Boycko et al. argue that divestiture will actually force a more rapid reform of local public finance, including the important step of privatizing commercial real estate. See Earle and Estrin (1995) for a detailed discussion of survey evidence on the structure of control rights in Russian firms, and implications for enterprise adjustment and restructuring. In a more complete framework one would want to model the market for benefits, but this is beyond the scope of the current paper. This follows from the fact that the elasticity of average product must be zero at n . The value a=.9 corresponds to a share of benefits in compensation (at zero subsidy) of about 10 per cent in the Cobb Douglas case a=l. The value of p describes the rate at which the output elasticity with respect to employment declines with the level of employment. Of course, this elasticity is unity at n’. The value p=-2.00 implies an output elasticity of 0.2 when there is a 50 p e r cent excess employment, n=l.5. We experimented with p between -1.5 and -2.0, corresponding to output elasticities (at n=1.5) of between 0.2 and 0.4. This range is consistent with available econometric evidence. See Milanovic (1995) and World Bank Russia Poverty Assessment (1995b). These numbers are taken from World Bank (1995b), pp.54-55. See World Bank (1995). The move toward cost recovery is accompanied by a programme of housing allowances to mitigate the income effects of rent adjustments. For a discussion of the implementation issues in this programme, see Raymond Struyk et al. (1996).


References

Estrin, Saul, Mark Schaffer and 1.J.Singh (1995), ‘The Provision of Social Benefits in State-owned, Privatized and Private Firms in Poland’, Centre for Economic Performance, London School of Economics, Discussion Paper No. 223, February.

Boycko, Maxim and Andrei Schleifer (1994), ‘The Russian Restructuring and Social Assets’, unpublished paper, May.

Boycko, Maxim, Andrei Schleifer and Robert Vishny (1995), Privatizing Russia, Cambridge, Mass.: MIT Press.

Commander, Simon, Une Lee and Andrei Tolstopiatenko (1996). ‘Social Benefits and the Russian Industrial Firm’, In: Commander S., Q, Fan and M. Schaffer, eds, Enterprise Restructuring and Economic Policy in Russia, Washington: World Bank, pp.52-86.

Commander, Simon and Andrei Tolstopiatenko (1996), ‘The Dynamics of Firm Restructuring’, unpublished paper, World Bank.

Commander, Simon, Qimiao Fan and Mark Schaffer, eds (1996), Enterprise Restructuring and Economic Policy in Russia, Washington: World Bank.

Commander, Simon and Richard Jackman (1993), ‘Providing Social Benefits in Russia: Redefining the Roles of Firms and Government’, World Bank Policy Research Working Paper No. 1 184, September.

European Bank for Reconstruction and Development (1995), Transition Report f 995, London, November.

Jackman, Richard ( 1 993, ‘Efficiency Considerations in Enterprise Provision of Social Benefits’, Economics of Transition, Vo1.3(2), pp.25 1-255.

Kaufmann, Daniel and Alexander Kaliberda (1995), ‘The Size of the Unofficial Economy’, Paper presented at University of Maryland and Johns Hopkins University

‘ Conference, Odessa. Le Houerou, Philippe (1999, Fiscal Management in the Russian Federation, Washington:

World Bank, 2 vols. Milanovic, Branko ( 1999, ‘Poverty, Inequality and Social Policy in Transition

Economies’, World Bank Research Paper No. 9, Transition Economics Division, April.

Schaffer, Mark (1995), ‘ Should we be concerned about the provision of social benefits by firms?’, Economics of Transition, Vo1.3(2), pp.247-250.

Struyk, Raymond, Alexander Puzanov and Lisa Lee (1996), ‘Monitoring Russia’s Experience with Housing Allowances’, unpublished paper, Urban Institute, March.

Tratch, Irina, Martin Rein and Andreas Worgotter (1996), ‘Social Asset Restructuring in Russian Enterprises: Results of a Survey in Selected Russian Regions’, in: The Changing Social Benefits in Russian Enterprises, OECD, pp.95-114.

World Bank (1995a), ‘Russian Federation: Towards Medium-Term Viability’, Country Economic Memorandum, Washington, DC.

World Bank (1995b), Russian Federation: Poverty Assessment, Washington, DC.

Date post:	21-Nov-2023
Category:	Documents
Upload:	inform
View:	1 times
Download:	0 times

Enterprise restructuring and social benefits

Documents