Determinants and effects of government size: Overview of theory and the Greek experience Chris Tsoukis Economics, London Metropolitan Business School Ministry of Economy, Competitiveness & Shipping Senior Research Fellow, Hellenic Observatory, LSE; January 2011
Outline International evidence on government/state size A look at Greece Overview of literature Social conflict and government size Model Findings Empirical implications
Evidence (Tanzi & Schuknecht, 2000) Secular rise in government size as a percentage of GDP in industrial economies Significant cross-country differences. Government expenditure/gdp: < 10% pre-world War I; about 15% shortly after; of the order of 30% in 1960; of the order of 50% in 1994; after which there may have been regressive tendencies).
Greece OECD data (general government unless stated otherwise)
Three sets of questions: Q. no. 1 (positive): What are the determinants of government size? Q. no. 2 (positive): What are the factors that determine the evolution across time of this government size? Q. no. 3 (normative): What are the appropriate economic borders and optimal size of the state?
Overview of Literature Economic development and country size as public services require a certain critical size, so that smaller economies tend to have bigger government (Alesina and Warciarg, 1998). Openness may be linked to government size in a variety of ways and with either a negative correlation (as in Alesina- Wacziarg) or positively because openness is a source of external destabilising shocks (Rodrik, 1996); Ethno-linguistic fractionalisation and instability (Annett, 2001); Modes of government and political representation - proportional and parliamentarian systems induce bigger government (Persson and Tabellini, 1999, 2004; Besley, 2004, 2005);
Overview of Literature (2) Productivity tends to raise the relative rise of government services which are not subject to much productivity growth (Baumol, 1967). Unemployment and government size may also be positively related (Nickell, 2004; Algan et al., 2002). Meltzer and Richard (1981): (purely) redistributive expenditure higher in more unequal societies (also Lipsett 1960; Prszeworski 2005). Dixit and Londregan(1996): opportunist politicians may target swing voters (as opposed to core ones), so spending related to the nature of political game, but unclear implications for government size.
Limitations No consensus. Unclear whether and why these factors might imply the evolution of government size mentioned above. Thus, the evolution mentioned above appears to be largely unresolved (see Drazen, Chapter 14); One seems justified to conclude that the determinants of government size are rather inadequately understood.
This work: Attempts to shed light on these questions (more the positive than the normative one). Question of government size embedded in a framework of social conflict. Extends previous work (Tsoukis and Tournemaine forthc g) which views the labour share (and therefore growth) as the product of social conflict and bargaining between two classes, capitalists and workers. The size of the government sector is also the object of bargaining, as it yields utility-enhancing services. Part of the bargaining over factor shares may now be supplemented by bargaining over government size. This framework yields a rich set of empirically testable insights on the determinants of government size.
This work -cont d. Not incompatible with the factors that were highlighted above. Key such determinant Bargaining power between capitalists and workers. Bargaining power Multifaceted concept: depending on labour and product market structures (unions, employment protection, welfare state, monopoly power in product markets, selfemployment/informal sector, corruption, etc.), structure of politics (parties, elections, parliament), policy (regulation), etc. This framework can potentially explain the evolution of government size, via (evolving) relative bargaining power of classes, as well as determinants.
Analytics 1 Capitalists (owners of firms) production function depends on private capital and growth-enhancing public services. y=τ Λ k Balanced budget, so rate of tax (τ) is also the size of government spending on goods and services. Workers consume their current wage bill. Capitalists are assumed to be dynamic optimisers whereas workers do not have the means of doing so.
Analytics 2 Rate of growth = incentive to invest (MRPK) impatience = (1-γ)(1- τ)τ Λ - ρ; γ: labour share Barro (1990) rule for optimal (growth-maximising) : w w k Workers utility: U = log ct + βτ yt + αct β: extent to which public services are also utilityenhancing; α: status motive or keeping up with Joneses c W =γ(1- τ)y= γ(1- τ)τ Λ k ( )
Analytics 3 k k Capitalists utility: U t = log ct + βτyt Without public services yielding any utility (β=0), both parties would choose Barro s τ* -as both growth and level of output involve (1- τ)τ Λ, which is maximised at τ= τ*. β>0 changes that: It creates an asymmetry, as this social income is not taxed (it involves only τ Λ ) and workers may depend more on that. Hence, workers are more inclined to support measures that enhance current income (τ Λ k) as opposed to growth (dependent on (1- τ)τ Λ. ( )
The indifference map and contract curve (CC) τ I W τ** CC τ*** I K γ
Implications 1 A rise in worker bargaining power will imply both a rise in labour share and a rise in the size of government (expenditure) particularly its utility-enhancing part. A rising βwill tilt the contract curve upwards. Any rise in worker bargaining power will therefore imply a faster rise in government size (perhaps less of an increase will be experienced by labour share). Thus, the rise in welfare state may increase government size directly (to provide services) but also because of this channel. An intensification of social comparisons (rise in α) shifts the contract curve up. The size of government (and the labour share) increases: As workers try to catch up and they cannot, they demand higher provision by government instead.
Implications 2 Rate of time preference: It shifts down the contract curve and decreases the size of government. Intuition: a more longtermist society (lower ρ) will want more government spending as this is growth-enhancing. More broadly, we should observe more long-looking societies (due to nature of political system, temperament, etc) having more growthenhancing expenditure (infrastructure, etc) whereas more short-termist ones having more welfare-enhancing spending (ceteris paribus). A final determinant (or whole group of determinants) of government size is the class game unions, political parties, their strength and organisation, nature of the game (democracy or not), nature of labour but also product markets (monopoly power, etc) regulatory framework, etc. Any increase of worker power should increase both government size and the labour share. Thus, labour share becomes a useful proxy for the relative power of the worker class (which needs to be instrumented by such measures).
Labour share by gov size (total expenditure) both % of GDP
Labour share by gov size (gov cons) both as % of GDP