Social Welfare and Productivity under Universal Social Rights in Mexico Santiago Levy,

1. Social Welfare and Productivity under Universal Social Rights in Mexico Santiago Levy, Inter-American Development Bank.

2. Background 1.) In Levy (2008) I argue that two aspects of social policy need to be considered simultaneously: • its impact on social indicators properly (how effectively are workers protected against risks, how effectively is income being redistributed, and so on); and, • its impact on incentives to workers and firms along dimensions that affect productivity and growth. 2.) Here I review the key arguments with a focus on the productivity aspect, and then present work-in-progress for a proposal to extend universal social rights to workers. 3.) I argue that this allows Mexico to escape from its present dilemma and increase equity and productivity. 4.) This issue is also relevant to poverty and the “post-CCT” discussion.

3. Labor market implications of social programs Total costs and benefits of salaried and non-salaried labor • Econometric estimates suggest that for unskilled workers: • Tax on salaried labor = 26% of the formal wage rate • Subsidy to non-salaried labor = 8% of the informal wage rate • The total tax-cum-subsidy to salaried vs. non-salaried unskilled labor implicit in Mexico’s social programs is in the order of 34%. This: (i) generates differences in the MPL between workers of similar characteristics and abilities, and (ii) distorts firms’ behavior along dimensions that hurt productivity (evasion and allocation of capital). • On the other hand, evidence shows that the minimum wage in Mexico is not binding, • and that there are barriers to entry only into a small sub-set of formal jobs, mostly in the public sector and some in very few private firms with monopoly power. There is very high formal-informal mobility.

4. Illegal behavior: evasion of regulations on salaried labor • When workers and firms hiring salaried employees have incentives to evade the Law. • Firms pay fines of F > Tf if they are caught evading. • Enforcement may be imperfect and there is a probability of being fined, , which is an increasing function of the level of evasion. • Evasion creates salaried workers without social security, Lif, receiving a wage wifthat compensates them for not receiving social security benefits. Although the illegal act is committed by the firm, both firms and workers may benefit from evasion. As a result of it, not all salaried workers are formal.

5. Minimum structure of the problem: • Firms hiring salaried workers maximize profits mixing Lf and Lif • Firms engaging with non-salaried workers maximize profits • Workers maximize utility and all are employed informal employment salaried employment Note: The equilibrium probability of being fined, , is determined endogenously along with wage rates [wf*, wi*, wif*] and employment levels [Lf*, Li*, Lif*]

6. Observed labor allocations and wage rates in Mexico reflect large tax-cum-subsidies associated with social policy θf Tf Lf= Subsidies to formal labor ≈ 0.6% of GDP (1-βf ) Tf Lf= De facto taxes paid on formal labor ≈ 2.4% of GDP (Lif + Li )Ti= Subsidiespaid to informal labor ≈ 2% of GDP (of which 0.65% of GDP are subsidies to workers hired illegally) This is the distribution of Mexico’s non-public labor force in 2006.

7. Policy needs an integrated view of social programs • It is essential to go beyond the effects of individual programs (components of Tf or of Ti) and consider their interaction. • Note that so the gap in marginal products increase with Ti. • One can show that but • so that the government is caught in a dilemma.

8. Firms have large incentives to engage in illegal behavior….. Steel or auto production are not profitable at small scales (relative to ), and therefore firms are always formal; but many activities in light manufacturing, services, commerce and transportation are. These are “informality prone” activities. • Depending on production technologies (particularly capital indivisibilities) firms • may find it profitable to be fully legal, semi-legal or fully illegal.The distribution • of firms across legal status is endogenous to social programs.

9. …… which matters a lot for productivity • firms mixing legal and illegal workers face increasing MCL so producing more output is more costly than for fully legal or fully illegal firms; • goods that are close substitutes or even the same goods are produced by firms of various sizes, with similar workers having different productivities. As a result: • the marginal productivity of salaried labor ranges from a low of wif to a high of (wf + Tf), so that there is a continuum of labor productivities; • the average of the marginal productivities of labor depends on the distribution of salaried workers across firms and on the size distribution of firms.The greater the proportion of workers employed in illegal or semi-illegal firms the lower is this average, and because the level of illegality of firms is inversely correlated with their size (measured by number of workers), the greater the proportion of micro and small firms, the lower is the average.

10. INEGI vs. IMSS registries of workers and firms, 2003 • Arguably75% of all firms in Mexico are illegal; this excludes economic • activity in the streets and in rural areas. Considering firms with 6 or more • workers, this share drops to 25%.

11. On average, TFP is positively correlated with firm size All manufacturing establishments captured in the 2004 economic Census (fixed premises) Source: IDB Productivity Flagship (2009).

12. Average TFP is a function of the size distribution of firms. The very large number of small firms is a cause of loss of TFP.

13. TFP dispersion in Mexico is much higher than in the US. If dispersion in Mexico was the same as in US, TFP in manufacturing would increase by 36%. Similar comparisons for services and retail commerce are not available, but gains would be presumably larger since dispersion of TFP in these sectors is larger than in manufacturing. Difference 90/10 percentile (in logs) Manufacturing 3.89 times Retail Commerce 4.90 times Services 4.15 times

14. Blue+ Red= urban Green= rural The Census Data fails to capture workers in urban areas that carry out activities outside a fixed establishment. Excluding public workers, about 45% of all urban workers are in this category……(and are at times missed in econometric analysis).

15. Can Social Policy Increase Welfare and Growth?

16. Policy needs to begin by recognizing that the government is trapped • The institutional distinction between the rights of salaried and non-salaried workers creates a formal-informal dichotomy; • From the social point of view, the government cannot leave workers excluded from social security without protection against social risks; • However, the instruments used do not really solve the social problem (contribution densities to pensions are low, workers are only protected against some risks when they are formal, and so on), while at the same time deepen the reasons that account for bad firms and bad jobs, low productivity and low growth; • In parallel, the government de facto subsidies illegal behavior, and undermines the tax base and the Rule of Law (as Lif workers cannot be left without social benefits).

17. The policy challenge has two parts: • From the social point of view, to insure that all workers are protected all the time against the relevant social risks regardless of whether they are salaried, self-employed, or comisionistas; • From the economic point of view, to provide benefits with programs that by-pass the distortions in the allocation of labor and capital and place the economy on a higher growth path with faster increases in labor productivity. • It is essential to focus on both objectives SIMULTANEOUSLY. (A big problem in Mexico is “piece-meal” policymaking.)

18. Proposal for universal social entitlements Simple idea: provide all workers with (almost) the same bundle of social benefits; make these benefits a legal entitlements; and fund all these benefits with a consumption tax earmarked for these benefits. Key point: Firms and workers cannot avoid this tax by changing status between salaried and non-salaried employment, so: • no taxes on firm growth; • no subsidies to self-employment; • no subsidies to small illegal firms; • no reasons to change the duration of labor contracts or disguise salaried employment relationships as non-salaried; • distortions in the allocation of capital and labor coming from social programs are (practically) eliminated. Observation: The result would be the same as if all workers were salaried and social security contributions were fully enforced. But since there are efficient reasons for non-salaried employment, and enforcement will never be perfect, universal social security will never be reached under the formal-informal dichotomy.

19. Social security contributions (A) vs. consumption taxes (B) If all workers were salaried, and if there was no evasion, and if consumption taxes could be ear-marked to pay for social security benefits, then social security contributions and consumption taxes would be equivalent.

20. Social Policy: Status-Quo vs. Core Proposal Status-Quo Core Proposal T* is the bundle of social benefits that the government considers that all workers should have (see below), and is worker’s valuation of that bundle; R*c reflects those consumption taxes that are specifically earmarked to pay for universal social benefits and that result from higher consumption tax rates, so that: T*(Lf + Li) = R*c.

21. Workers’ utility: status quo vs. core proposal without redistribution Under the core proposal the government’s objectives are fulfilled with all workers but worker’s utility may or not be higher. + ? -

22. Workers’ utility: status quo vs. core proposal with redistribution To achieve both the objective of changing the composition of all worker’s consumption and increasing the level of some worker’s consumption you need two instruments. The proposal is to Slutsky-compensate (or more) a subset of workers with direct income transfers, choosing s* such that:

23. How much would universal entitlements cost? Estimated Net Fiscal Costs of Universal Social Entitlements (millions of pesos of 2007) Dt*c= 4.3% GDP But was 2.7% and now is 0, so net Dt*c= 1.6% GDP The proposal implies increasing the VAT and (almost) eliminating labor taxes: a large change in the composition of taxation from labor to consumption.

24. Social security and social protection programs impact the budget on both sides Resources for social security and social protection programs for active workers • It is important to separate social spending on social security and social protection programs as defined from other social spending. • As Ti increases there are more pressures on the budget, since incentives to firm’s illegal behavior are higher (and profit taxes lower), while the number of informal workers receiving free social benefits increases. • The question is how the reform changes the budget constraint of the government as output levels and the incentives to evade change . Resources for education, targeted poverty programs, and other social spending. Consumption taxes depend on real wages and evasion of VAT Profit taxes depend on the number of firms that are formal and evasion behavior

25. Full de-linking and full un-bundling is not desirable Structure of benefitsObservations 1. Income transfers s* same amount all workers Workers in households up to the third decile receive a net income transfer 2. Re-design of social security T* = [health insurance retirement pensions Bundled protection for all workers for these risks life insurance disability insurance] Tf Tf* = [work-risk insurance unemployment Salaried workers get additional protection for insurance (extra retirement pensions?)] risks specific to salaried work Other = [housing loans + day care centers benefits + sports and cultural facilities] These benefits are not legal entitlements Labor Market Firms hiring salaried workers pay for Tf* ( 4%) Non-salaried workers do not get Tf* Wage-based benefits distort only if ; note that Tf* contains only monetary benefits, so Note: This scheme allows for a two-pillar retirement pension system (only when workers are salaried-employed), by including a retirement pension component in Tf*.

26. A Digression: The VAT in Mexico

27. Note that: • In 2006 VAT revenues in Mexico were only 4.2% of GDP, with an average VAT rate of around 8% because of many exemptions on food, medicines and the like justified on redistributive grounds; • The proposal takes advantage of Mexico’s extremely high unequal distribution of income and consumption: out of every peso of additional revenues from VAT only 10 cents are required to Slutsky-compensate the first two deciles and 27 cents the first five deciles (Davila and Levy, 2003); • The proposal implies setting a VAT rate of 15% with no exceptions, and compensating all workers with a direct income transfer that leaves households in the first two deciles of the distribution with the same real income as before.

28. 100% consumption 6.9 11.5 15.4 21.1 45.0 100 70.9% of GDP 90 80 4.9 8.2 10.9 15.0 31.9 70 % share of monetary consumption 60 50 40 30 20 10 0 20 40 60 80 100 % share of households The distribution of consumption is very skewed in Mexico The poorest four deciles account for less than 18% of all consumption, while the richest four deciles represent more than 66%, according to the 2006 income-expenditure survey.

29. 11000 10000 6.5% del PIB 9000 8000 0.48 1.05 1.40 2.74 0.80 7000 Additional consumption taxes (pesos per month) 6000 5000 4000 3000 2000 1000 0 20 40 60 80 100 Percentile of households The VAT is a good instrument to raise revenues in Mexico. The proposal is to tax all goods at the same 15% rate Only 7.3% of the additional revenues would come from the first two deciles, and 20% from the first four deciles, while the richest four deciles would pay 64%.

30. -0.48 -0.80 -1.05 -1.40 -2.74 -6.5 +0.48 +0.46 +0.47 +0.46 +0.44 +2.3 +0.89 +0.84 +0.87 +0.85 +0.74 +4.2 +0.89 +0.50 +0.29 +0.2 -1.36 -0.0 Recaudación adicional del IVA 6.5% del PIB Beneficios sociales universales adicionales 4.2% del PIB* Transferencia Monetaria por compensación 2.3% del PIB** Additional VAT collections are fully returned to workers at the level of the second decile * Equivalente a 800 pesos mensuales por trabajador ** Equivalente a 435 pesos mensuales por trabajador

31. Redistributive impact (Poorest) (Richest)

32. A note on productivity and poverty • This line of work is relevant to the “post CCT” discussion. • In Levy (2008) I argue that and that and show that this helps to explain this puzzle: • why, if most poor workers have no assets other than their labor, they end up in informal jobs, when the expectation would be that they should have formal jobs? • There is an incentive-compatibility problem between Progresa, on one hand, and social protection and social security programs, on the other.

33. D P P P P P,NP NP,P a ) L = L – R + L – L + L t t t t t +1 D NP NP NP NP P,NP NP,P b ) L = L – R + L + L - L t t t t t +1 Poor workers future capabilities Incentives to firms and workers Outcomes in the labor market Social security and social protection P P U = U programs distort firm’s and workers f i Progresa - Oportunidades decisions in the labor market NP NP U = U f i subsidizes the demand for health and education but of poor children and youngsters MPL > MPL b b b b P NP P NP < ; > f i f f i i and P NP P NP T > T ; T = T i i f f P P NP NP (L /L ) < (L /L ) f f D f , i i,f c ) L = L – R + L – L + L t t t t t f +1 f f f D f,i i,f d ) L = L – R + L + L - L t t t t t i +1 i i i Progresa cannot fix the problems created by the formal-informal dichotomy. Raising benefits in Progresa indefinitely cannot substitute for a job with higher labor productivity. And raising Ti to increase the welfare of poor workers in a context of stagnant real wages makes the productivity problem worse.

34. Work in progress (my KCP): • Study the incentives to evade VAT under the proposal, comparing a VAT reform on its own with a (VAT + social) reform; • Study firms’ incentives to evade income taxes, again in the two scenarios; • Study an unemployment insurance that has lower efficiency costs than the current system of severance pay and improves benefits to workers; and, • Consider the productivity and the social welfare side at the same time, but think of all these themes as “pieces of a puzzle” and not in isolation. Thank you.