Professor Jim Stewart:

I thank the sub-committee for inviting me today to assist it in relation to measuring corporate tax rates. I have been engaged in researching the area of effective tax rates over a considerable time period. My initial research was conducted for a PhD thesis submitted to the London School of Economics in 1985 and subsequently published as a book. I would like to briefly describe my research approach because it illustrates a number of issues that arise in measuring effective tax rates. First, the concept itself is relatively simple. The tax rate paid on corporate profits for an individual firm can be defined as tax paid divided by pre-tax profits. When we try to measure this empirically, abiguities immediately arise.

The tables included in my circulated opening statement include an illustration of the foregoing in respect of an Irish public company. Table 1 shows pre-tax profits and other data for the most recent three-year period and also shows three possible measures of effective tax rates. Measure 1 shows the effective tax rate defined as the tax charge in the profit and loss account divided by pre-tax profits, again taken from the profit and loss account. This measure is described in company accounts as "the effective tax rate". Measure 2 shows the tax actually paid divided by pre-tax profits. Cash actually paid is set out in the cash flow statement. When I did my research initially, one had to calculate this, but it is now regularly produced in all cash flow statements of public companies. The aggregate of tax actually paid is the number shown in Revenue statistics for corporate tax receipts in any given year. If one aggregates that across the entire corporate sector, one should have the number shown in Revenue statistics, allowing for timing differences. Measure 3 is similar to measure 2, but includes provision for accounting depreciation in the tax base - the denominator. The reason for this is that accounting depreciation in contrast to other measures of depreciation such as accelerated depreciation is not a tax deductible expense. We have a wider definition of the tax base and could make our other adjustments as well.

Table 1 sets out the difference in the effective tax rate for 2013 across the three measures being 12.5% under measure 1, 4% under measure 2 and 2.7% under measure 3. Measure 3 is often used in academic studies to estimate effective tax rates while measure 2 appears to be the measure of individual tax rates used in studies by the US Permanent Subcommittee on Investigations, or PSI, in its various reports. In almost all cases of which I am aware, the tax paid in the cash flow statements is less than the tax charge.

Broad-based measures of effective tax rates may be estimated by first of all calculating individual tax rates for each firm. Measures of effective tax rates for the economy as a whole can then be calculated by summing the data for individual companies across the entire economy. This could be done by treating each firm as a separate observation and equally weighted, and calculating means and other distribution statistics. In addition, averages could be calculated by aggregating data across all firms. In this latter case, average effective tax rates will be dominated by firms with the largest profits and tax rates. The bigger the firm, the greater the weighting it has in the aggregate data. US Bureau of Economic Analysis data uses this latter approach.

In an earlier study which I referred to, I estimated effective tax rates over the 21 year period 1964-1984 for all Irsh non-financial plcs treating each firm as an individual observation and then aggregating them in various ways. However anomalies can arise. Rather than paying corporation tax, a firm may receive corporation tax because of overpayments in previous years. Hence the numerator will be negative resulting in a negative effective tax rate. A firm could both receive tax payments in the current year and report losses in the current year. In this case, both numerator and denominator are negative, -T/-P, resulting in a positive effective tax rate. A firm could pay tax even though losses were reported. This could arise because corporate tax payments are paid in arrears. Hence, the measure of effective tax rates is negative. This is less likely to occur than in previous years because lags in corporate tax payments have been reduced. A firm could report small positive tax payments but report low profits, resulting in a very high measured effective tax rates. In large data sets all these problems arise. One solution is to estimate effective tax rates over a running three year period. However, anomalies will still remain. A more usual approach is to estimate effective tax rates for only those firms reporting positive profits.

It is likely that these issues are endemic to Irish data. Table 2 in my presentation shows that for 2011, 1% of firms - 433 firms - account for 81.1% of corporate tax payments. It is quite likely in fact that 50 firms account for in excess of 50% of total corporate tax payments. I note that 76.8% of all firms account for 3.56% of corporate tax payments and 38.8% of firms report zero or negative net trading income. Reporting zero or negative net trading income is not due to having zero taxable profits and tax payments on foot of being able to afford expensive accounting and tax advice, but because companies are making losses. We have come through a very severe financial crisis. Firms that report losses in one year are likely to report losses in succeeding years. They may eventually close. Finally, reported losses by banks will have had a major impact on corporate profitability in Ireland and measures of effective tax rates.

I will say something briefly on aggregate data. National accounts in Ireland are currently based on what is described as the income and expenditure approach and in future an output approach will be incorporated into the GDP compilation process. It will be interesting to see what change that makes on the big numbers. Aggregate data, such as profit estimates from the national accounts, ignores issues relating to the inclusion of firms making losses. What one really wants is a measure of the effective tax rate on those firms making profits. Aggregate data cannot be interpreted as indicating effective tax rates for the median or typical firm. For this reason, empirical studies of measures of effective tax rates in the finance, accounting and taxation literature use data from accounting records - either publicly available or from surveys of firms which report their own accounting records as in the US Bureau of Economic Analysis data - rather than aggregate data contained in the national accounts. I found in an earlier study of US Bureau of Economic Analysis data that the effective tax rates using such data on one measure was 2.2% while it was 3.8% on another measure. That is for US companies incorporated in Ireland.

In addition in Ireland, particular issues arise from the likely underestimation of profits arising in the IFSC. The IFSC had over €3 trillion in gross assets in 2011. If we assume a net rate of return net of all expenses of 1%, that is €20 billion. However, it is very difficult to find those kinds of numbers in the national accounts. The exclusion of those firms that are incorporated in Ireland but regarded as resident for corporate tax purposes in another jurisdiction poses another difficulty in estimating effective tax rates using aggregate data.

I turn to the issue of company location. The Department of Finance argues that these firms should be excluded from Irish data as they are not located within Ireland for corporate tax purposes albeit that they are incorporated here. If a firm is incorporated in Ireland, the auditor must be located in Ireland by law. If a company is registered in Ireland, the books of account must be maintained in Ireland. Several of these companies that claim not to be tax resident in Ireland have employees here and hence are liable to pay employers' PRSI. It is not then true that they are exempt in Ireland because they are tax resident in Bermuda. If they have employees here, they must pay PRSI. Several of these companies are liable for VAT and hence satisfy the requirements of residence for VAT purposes. I acknowledge that those residence requirements are different from those for corporation tax purposes. The European Court of Justice has ruled that the place of incorporation is key to where the central management and control exists. Place of incorporation is therefore very important for corporate law purposes. EUROSTAT regards brass plate firms with no fixed assets or employees as being located where they are incorporated, a rule which is likely to be followed by the Revenue Commissioners regarding the various brass plate firms located in the IFSC.

In conclusion, estimating effective tax rates is best done using data extracted from company accounts. Estimating effective tax rates from national accounts is likely to produce an underestimate of effective tax rates because of the inclusion of firms making losses as well as for other reasons.