Thursday, January 24, 2013

The Tax/GDP ratio and Mr. Subramanian's strange contention

Arvind Subramanian, has written a curious op-ed yesterday in which he tries to judge the performance of the state of Gujarat vis a vis other states through its ability to generate tax revenues as a % of its GDP. This is a slightly unconventional metric, and Mr. Subramanian lays out the philosophical justification for it early on. He says
Historically, institutional development has been associated with effective taxation. Taxation is the glue that binds the rulers and the ruled; fact that they are taxed gives citizens an incentive to hold rulers accountable, and this in turn helps to ensure that the latter deliver value for money by way of efficient services to citizens. “No representation without taxation” is, thus, one of the key insights of political and economic development. A corollary is that reasonable (but not onerous) taxation ensures that good governance will be durable, outlasting individual leaders
This sounds reasonable in so far as it goes. Mr. Subramanian makes both a cross-sectional as well as longitudnal comparison, i.e. between states as well as across time. Globally, cross-section tax/GDP ratios have a debatable relation with state capacity but the time-series shows the Wagner law - increasing prosperity is strongly correlated with an increasing tax/GDP ratio in most nations. So far so good.

But the leap that Mr. Subramanian makes from there, into positing that the own tax revenues (OTR)/ gross state domestic product (GSDP) ratio for Indian states is an operationalization that accurately captures the letter and the spirit of the above philosophy and empiricism, is unwarranted. India's highly centralized fiscal framework means that states collect only a limited set of the taxes that its consolidated governments impose on its citizens and firms. Broadly, these include goods/consumption taxes (state VAT and excise), stamp duties, registration taxes, motor vehicle taxes, tolls, etc. Income taxes, corporation taxes, central excise, service taxes etc. are controlled by the centre and then evolved to the states through finance commission recommendations. It's not at all clear why a taxation schema that is entirely devoid of direct taxes can be considered a good guide to the fiscal or development capacity of a particular state government in India. In other words, even if one accepts Mr Subramanian's philosophy, his key metric does not necessarily follow.

So what does the OTR/GSDP represent? Mr. Subramanian subjects it to a smell test - we 'know' that AP, Karnataka, TN are governed well and they seem to have high OTR/GSDP and we 'know' that West Bengal is governed poorly, and it has low OTR/GSDP, so surely we must have picked the right metric. But this smell test raises more questions than it answers.

1) Good governance was supposed to be important for economic development - where is the proxy for economic development? Is it per capita GSDP and we're just supposed to 'know' it?

2) Even so, why would a schedule of indirect taxes and stamp duties that are not progressive in rates (for a given state, Rs.100 spent on a certain commodity or transaction attracts the same rate of tax as Rs. 10,000, unlike, say, for income taxes) display progressive properties and be increasing in economic development?

3) If consumption taxes in India are indeed highly progressive through qualitative differences - e.g. essential products are charged at 0%, while non essentials variably between 4%-5% and 12.5% - do the differing tax/GDP ratios simply reflect differential consumption mixes across states rather than any governance attributes? Does this have development or state capacity implications?

4) Even beyond a general progressive effect in consumption, do the differential OTR rates simply indicate one or two critically divergent sources of sales and excise revenues? To give one example, the states of AP, TN & Karnataka all collect own tax revenues of the order of Rs. 400 billion every year, of which nearly 50 billion each year comes from liquor sales, which are banned in Gujarat. This is about 1% of the GSDPs of these states. Is that all there is to the inter-state comparisons? Does a divergence between states in liquor revenues over the last 20 years mostly explain the time-series divergences?

5) If the consumption patterns aren't very different, does the OTR/GSDP simply reflect differential tax rates for similar commodities or transactions? If so, does it show up in the price levels between the states and has this been controlled for? Should it be controlled for? Do inter-state price differences have consumption implications, controlling for wealth and income?

It is far from obvious that OTR/GSDP is a measure of anything that proxies well for the development capacity of a state.

Mr. Subramanian is upfront about some challenges that he sees could be offered to his thesis, including what he describes as the ideological right-of-centre argument of low taxes = private sector led growth. Surprisingly, he offers no counterpoint of his own to this challenge - having seen the opponent's move, why does he not think ahead and play his next one? But this line of argument is not central in my view, given that direct taxes on personal income and corporation income have simply been excluded from the states' tax schema and thus this analysis by construction.

A typical critique of Mr Subramanian's has been made by R Jagannathan at Firstpost. In his usual style he packs one or two good arguments with a set of weak and annoying assertions. Mr Jagannathan rightly takes Mr Subramanian to task for speculating on Gujarat's development performance as a function of its OTR/GSDP ratio without considering Kerala, typically offered as the anti-dote to the 'Gujarat model'.

However, he goes on to make some rather strange assertions. He talks about Gujarat's agricultural spike that yields no income taxes, but this is neither here nor there - an explosion of skilled salaried professionals in Bangalore will similarly increase Karnataka's GSDP but not so much its OTR because they pay the their income taxes to the centre, not the state.  He argues that stable and increasing tax/GSDP ratios are a function of political stability. Sorry, what? Political stability helps tax revenues grow as a % of GDP how exactly? Can we atleast have a bare-bones argument to place our faith upon? He also seems to believe that since Gujarat focuses on infrastructure and the other states on services, which are inherently 'profitable', the other states have an advantage in revenue collection. Again, how exactly? What taxes do these non-infrastructure corporations pay to their respective state governments out of their outstanding profits? He goes on to allege Gujarat's neighbouring states for its failure to draw more revenues from infrastructure. Umm, ok. These are shallow speculations, excusable in a reactive casual conversation, but not worth a place in what is supposed to be an analytically founded editorial response. (To be fair to him, the greater disappointment is that Mr Subramanian chose to make this about Narendra Modi's performance)

Another more interesting challenge is further offered by Mr Subramanian himself - again surprisingly without a counter-point of his own - where he sees Gujarat as offering a different set of tax vs service trade-offs to its inhabitants than other states, akin to a differentiating strategy in the classic Tiebout model of competitive taxation. But we come back to the same question - what really are the tax benefits being offered by the state of Gujarat to lure private investors that are differential? Stamp duty removal, capital returns for investments over a certain limit, exemption of electricity duties for SEZ units - these are bog standard sops offered by all industrial states to investors. Compare, for example, Gujarat's industrial policy strategy with Karnataka's and tell me if they're not both bending over backwards to exempt stamp duties, offer interest subsidies, re-imburse this duty and that. Not so long ago, we had articles arguing that other states may offer more sops, but Gujarat's development model was more sustainable. No, fiscal competition does not quite cut it.

Ultimately, we're left with the same fundamental dilemmas on the use of OTR/GSDP that we started with:
1) Given that ALL direct taxes are excluded, is the pure theory of taxation and state capacity even applicable to the OTR/GSDP ratio?
2) What is the underlying model and cause of progressivity here? OTR/GSDP is progressive as a function of what exactly? Is that something a good proxy for of a state's fiscal capacity?

A couple of other, minor points are also in order. It doesn't help that Mr. Subramanian words a key sentence rather poorly. Speaking of Gujarat's decline in the OTRGSDP ratio over the last 20 odd years, he says:
This decline is actually dramatic — because this was a period during which Gujarat grew very rapidly, which should have elevated its tax collections
As can be expected, this statement can be read as a schoolboy mistake, confusing absolute numbers and percentages, which I don't think he's guilty of. He's most likely making a point conditioned on the assumption of progressivity of OTR/GSDP, which goes horribly wrong exactly because he has not elsewhere given evidence that he has handled the question of progressivity.

Further, the data that he offers in his charts and tables - though I think they're only tangential to the key issues here - look somewhat strange. The average OTR/GSDP seems overstated by more than a percentage point, going by this data from RBI which is referred to as a source by Mr Subramanian's charts. Perhaps I'm looking at the wrong set of figures - the main RBI report is close to 400 pgs and I've not had the chance to go through it in detail. Still it would have been nice to be able to easily reproduce Mr Subramanian's data at source.

To conclude, it is unfortunate that Mr. Subramanian chose to use his preliminary findings to make what is a weak and unfounded political point, because the pure analytics of the line of research he's pursuing is important, interesting and prima-facie, deleteriously stunted. The business of comparing this chief minister with that and speculating on comparative development performance with one unsettled metric is best left to political dilettantes, available a dime a dozen in the edit pages of our main broadsheets. I would have expected Mr Subramanian to offer insight, and engage in a deeper meditation on refining precisely what measure of fiscal capacity might indicate development ability, and why. His academic work on this with Utsav Kumar is still forthcoming, and one can only hope that in its more developed form, it will grapple honestly with the sort of questions that I've raised here.


Zagen Germaine said...

We will be in the 28% tax brackets in 2013. I'm happy because we are skating close to the edge and the 2014 changes will keep us in the same tax bracket.

Infinity said...

Arguments of the blogger are complex in terms of language and shallow in terms of logic. For example: He denies direct relation between political stability and tax/gdp ratio.... does he understand that political stability prevents bureaucratic anarchy and corruption hand in glove with traders and businessmen thus higher tax collection. And he is not able to see that though state doesn't get income tax but higher income in service sector cause higher spending and demands thus larger indirect tax collection. I wished that blogger didn't implicate the reader in incoherent language and explained things in simpler way....writing complex sentences and thus obscuring simplicity won't solve the purpose for which blogger writes tthe blog I.e to make others understand.