VoxEU Column Europe's nations and regions Frontiers of economic research Global crisis

Trying to glimpse the ‘grey economy’

Despite the growth of online and card payments, the ratio of currency to GDP in the UK has been rising. This column argues that rapid growth in the grey economy has been a key cause. The authors estimate that the grey economy in the UK could have expanded by around 3% of UK GDP since the beginning of the Global Crisis.

It is a remarkable fact that the ratio of currency to GDP in the UK has been rising, despite the greater use of card and online payments (see Figure 1). The currency-to-GDP ratio now stands at 16.1%, compared to 13.3% in Q4 2007. Currency in circulation per adult person is now equal to around £1,300 in the UK. In some other equivalent cases – e.g. holdings of US dollars, euros, and Swiss francs – this might be because more currency is being hoarded abroad, but this is not likely to be the case for the UK. So what is fuelling this rise in currency usage, if not the grey economy?

There has been ample coverage in the British media about the upcoming inclusion of illegal activities such as prostitution and drug dealing into the UK National Accounts, which should boost measured GDP by around 0.7% (see Appendix). More important, if it could be done, would be an attempt by the statistical authorities to provide a plausible estimate of the size of the ‘grey economy’. The currency-to-GDP ratio has been expanding rapidly since the onset of the Global Crisis (see Figure 1), and we believe rapid growth in the grey economy has been a key cause. Using the ‘currency demand’ approach, we tentatively estimate that the size of the grey economy has expanded by around 3% of UK GDP since late 2007. There remains, however, material uncertainty around such an estimate.

Figure 1. The currency-to-GDP ratio has soared in recent years

Sources: Office for National Statistics (ONS), Bank of England, Morgan Stanley Research

The grey economy, defined as otherwise-legal activities that are deliberately not recorded in order to avoid or to evade taxation, has almost certainly been expanding, and probably quite fast. Why do we think that? Of course we cannot be sure, because the grey economy’s rationale is largely to evade tax – especially VAT, National Insurance (social security) contributions, and Pay As You Earn (income tax) contributions – and hence goes unreported. The easiest way to avoid records of taxable transactions is to mediate these via cash payments.

Estimating the growth of the grey economy

We use the currency demand approach to estimate the likely increase in the size of the grey economy. The currency demand approach has historically been one of the main methods for estimating the size of the ‘shadow economy’ (grey plus black economy).1 The basic idea is that, since tax evasion is illegal, almost all grey (and black) economy transactions will be made in cash.2 For obvious reasons, cash is almost always anonymous, whereas most other payment mechanisms leave a record. What one does, then, is to estimate how much of the currency-to-GDP ratio is due to incomes, interest rates, technological trends, and such other variables as theory or direct observation suggest (a standard currency demand regression). One can then either take the residuals from such an equation as an estimate of the shifting shape of the hidden economy, or, better, add additional variables that should be correlated with the grey economy, such as tax rates – especially VAT – and the ratio of the self-employed and unemployed to the total workforce. We did the latter.3

There are, however, a couple of other factors that probably have raised currency holdings in recent years. The first is the decline in interest rates to nearly zero. If you cannot earn interest on a bank account, there is less incentive to put spare cash on deposit in a bank. We test for this effect in our empirical exercises and, like most other studies of this kind, we find that the demand for cash holdings is reduced if interest rates on bank deposits rise and vice versa (i.e. that the interest elasticity of cash holdings is negative), although the impact is relatively slight in terms of magnitude. (In this particular instance, however, the decline in interest rates was very large, meaning the overall effect was quite sizeable).

Of more interest is the jump in cash holdings starting in early 2008 (see Figure 1), in the aftermath of the Northern Rock nationalisation and the subsequent Lehman bankruptcy and RBS/HBoS rescues. These events alerted depositors to the potential fragility of banks, and we believe that the cash-holdings jump can be largely ascribed to public worries about the security and future availability of their deposits. But such concerns were assuaged by the successful measures taken by the authorities to defuse the Global Crisis. However, the cash ratio did not fall back after 2009/10, once the Crisis had been defanged. Was this because other, notably grey economy, factors took over, or perhaps because there remained some residual concern over the safety of bank deposits? Frankly, we do not know for certain, but our belief is that confidence in the safety of bank deposits was, more or less, fully restored within a couple of years, and our preferred econometric equation incorporates that assumption. We plan to do more work on the effect of the Crisis on the demand for currency.

There is a general belief, which we share, that the self-employed find it easier to operate in the grey economy than those in employment. Some individuals may prefer to be in self-employment partly because of the subsequent ease of avoiding, or evading, tax on incomes and/or consumption. Certainly, as reported in a previous paper of ours (Ashworth et al. 2014), self-employment has grown strongly in the UK in recent years (see Figure 2), and much more so than in other developed countries (see Figure 3). This would seem to be consistent with a fast-growing grey economy, and this remains our own belief. Yet despite considerable effort, we have failed to find econometric support for this hypothesis in our empirical work. We used as our main variable the ratio of self-employed to total employment (SETE). As shown below, this rose initially in the early 1990s (see Figure 4), then fell from 1995 until the bursting of the technology bubble, before rising again in recent years. While the recent rise coincides with the rising cash ratio, the direction of movement prior to 2008 was not typically positively correlated. We could not find a significant relationship between SETE and the currency-to-GDP ratio. While we still believe that there will be some relationship between an increase in self-employment and a larger grey economy, this issue may be more complex than we had initially expected.

Figure 2. UK self-employment, millions

Sources: ONS, Morgan Stanley Research

Figure 3. Self-employment has fallen in most other countries since end-2007

Sources: Haver Analytics, Morgan Stanley Research

Figure 4. UK self-employment as a percentage of total employment

Sources: ONS, Morgan Stanley Research

We have enhanced our previous attempt to model the grey economy. One of the problems with our previous empirical exercise (Ashworth et al. 2014) was our assumption that the response of the currency-to-GDP ratio to our variables was immediate, rather than building up over time.4

What we hoped to find is a metric that can serve as a proxy for a shift of expenditures out of the recorded economy into the grey economy – and we may have found one such metric. Unlike house-building, which will be fully recorded, minor housing repairs are the sort of activity that could slip into the grey economy. One might start from the default assumption that, were all transactions fully and correctly measured, housing repairs would represent a constant proportion of the value of the housing stock (a similar assumption is made about depreciation), although there are, admittedly, several other reasons why this ratio could vary. Then a decline in the ratio of house repairs to the housing stock might become a proxy for a rise in the grey economy. This ratio has declined quite rapidly over recent decades. It has stabilised more recently, however, perhaps as the average age of the housing stock has increased.

Our variables for seeking to explain the recent rise in the ratio of currency holdings to GDP are now as follows. These are in log form:

i) Lagged dependent variable;

ii) GDP, to account for various (technological) trends and income elasticity;5

iii) Housing repairs as a percentage of the housing stock;6

iv) Lagged VAT rate;

v) Interest rate;

vi) Three dummies for the 2007–2009 spike due to the Crisis, with the final one preventing any subsequent, longer-lasting effect.

The resulting regression equation, shown in Table 1 below, may look splendid, with an adjusted R2 of over 0.99, all coefficients with the predicted sign and highly significant (except GDP), and no autocorrelation among the residuals. But beware. The equation was over-fitted, data mining was extensive, and no out-of-sample tests have been done. The equation could easily fall apart in the face of future developments. All that it really tells us is that the suspicious rise in the currency-to-GDP ratio could have largely come from a rise in the grey economy.

Table 1. Explaining the recent rise in the currency-to-GDP ratio

Note: We also included three dummies to deal with the effects of the financial crisis.

If, however, we should take this equation at face value, then the effects of the two grey economy variables – the VAT rate and housing repairs as a percentage of the value of the housing stock – would appear to account eventually (i.e. after taking account of the interaction with the lagged dependent variable) for around half, or about 10 percentage points, of the rise in the currency-to-GDP ratio of around 21% since 2007. (The VAT variable was the primary driver.) This is less than our preliminary estimate of 20% (Ashworth et al. 2014).

There is then a further major problem of estimating what proportion of currency outstanding is held to facilitate the grey economy. Recall that the value outstanding of currency per UK adult amounts to around £1,300 per head. To reach that level, one surely must assume that a sizeable proportion is held for black or grey economy purposes, and we believe that the black economy has been contracting recently (see Appendix). If we should assume that around two-fifths (40%) is held for such purposes, then a rise of 10% in currency holdings would translate into an increase of about 25% in the grey economy (more if you think that 40% is too much). And if the shadow economy (the bulk of which, in our opinion, is the grey rather than black economy) initially represented about 12% of the total UK economy, it would by now have grown to 15% (almost entirely due to growth in the grey economy). In other words, we think that the underlying economy may have grown since 2007 by almost 3% more than the recorded ONS data would suggest.

Conclusion

We estimate that the grey economy in the UK could have expanded by around 3% of UK GDP since the beginning of the Global Crisis. Admittedly, our analysis involves a surfeit of ‘ifs’, piling uncertainty upon speculation. Nevertheless, a reasonably accurate guesstimate of the grey economy is relevant when addressing a number of key policy debates. Indeed, the numbers are potentially large and could influence our perception of the UK economy, the productivity puzzle, and much else besides. For this reason, we would strongly welcome more resources being provided to the ONS to aid it in its efforts to measure the magnitude of the grey economy.

Charles Goodhart is a senior adviser to Morgan Stanley. In his personal capacity, Mr Goodhart advises other organisations and firms on economic matters, including, among others, the Financial Markets Group of the London School of Economics.

Authors' note: This article is based on research published for Morgan Stanley Research on 7 April 2014 and 3 September 2014. It is not an offer to buy or sell any security/instruments or to participate in a trading strategy. For important disclosures as of the date of the publication of the research, please refer to the original pieces available at UK Economics: The UK’s self-employment phenomenon: why the labour market isn’t so strong after all (7 Apr 2014) and United Kingdom: Trying to Glimpse the ‘Grey Economy’ (3 Sep 2014). For important current disclosures that pertain to Morgan Stanley, please refer to the disclosures regarding the issuer(s) that are the subject of this article on Morgan Stanley’s disclosure website (https://www.morganstanley.com/researchdisclosures).

References

Ashworth, J, C A E Goodhart, and M Baker (2014), “The UK’s self-employment phenomenon: why the labour market isn’t so strong after all”, Morgan Stanley Research, available on request from authors.

Cagan, P (1958), “The demand for currency relative to the total money supply”, Journal of Political Economy 66: 302–328.

Drehmann, M and C A E Goodhart (2000), “Is Cash Becoming Technologically Outmoded? Or Does it Remain Necessary to Facilitate Bad Behaviour? An Empirical Investigation into the Determinants of Cash Holdings”, LSE Financial Markets Group Discussion Paper 358. 

Isachsen, A J and S Strom (1985), “The Size and Growth of the Hidden Economy in Norway”, Review of Income and Wealth, 31(1): 21–28.

Schneider, F and A Buehn (2013), “Estimating the Size of the Shadow Economy: Methods, Problems and Open Questions”, CESifo Working Paper 4448. 

Tanzi, V (1980), “The underground economy in the United States: estimates and implications”, Banca Nazionale del Lavoro Quarterly Review 135: 427–453.

Tanzi, V (1983), “The underground economy in the United States: annual estimates, 1930–1980”, IMF Staff Papers 30: 283–305. 

Appendix

The only type of illegal activity currently covered in the UK National Accounts is the smuggling of alcohol and tobacco. However, with the release of the National Accounts data consistent with Blue Book 2014 on 30 September, the ONS will for the first time include prostitution and illegal drugs activity in its measurements of the economy’s output.7 According to the ONS, the inclusions of these ‘black economy’ activities will boost measured UK GDP by around 0.7%.

We suspect that the size of the black economy has actually been declining for some time though. Reported crime and drug usage have been declining for a number of years (see Figure 5). This measure, however, excludes cyber crime, which has been rising. It is not clear, however, how many UK resources are used up in this process. The number of recorded offences in England and Wales has declined by 60% since the peak in 1995 and has continued to decline in recent years despite the deep and extended recession in the UK.

Figure 5. Estimated crimes in England & Wales

Source: Crime Survey for England & Wales, ONS

Footnotes

1 See in particular Schneider and Buehn (2013), especially Section 3.2.4.

2 Not all transactions in the shadow economy are paid in cash. Isachsen and Strom (1985) used the survey method to find out that in Norway, in 1980, roughly 80% of all transactions in the hidden sector were paid in cash.

3 This has been one of the most commonly used approaches, first used by Cagan (1958), extended by Tanzi (1980, 1983), and later employed by one of us (Drehmann and Goodhart 2000). More recently, however, it has fallen out of favour. Perhaps the main reason for this has been the scale of the international use of the US dollar, and more recently of the euro, both for transactions and savings purposes. This makes it hard to disentangle in those instances domestic from international determinants of the demand for currency. Fortunately for our purposes, pound notes are not unduly held, or usable, abroad. Moreover, black (illegal) transactions are largely undertaken in high-denomination notes (such as €500 or CHF 1,000), and the highest-denomination UK note (£50) is commendably low value. The £50 note was introduced on 20 March 1981, and is a low-denomination note compared to those available in the euro and Swiss franc.

4 In addition, the residuals from the equation were auto-correlated as shown by the low Durbin–Watson statistic – a result which generally implies that the functional equation is in some way misspecified. Both these problems could be addressed, in part, by including a lagged dependent variable, though running such an equation with dummies (for the Global Crisis concern about deposit safety) is complicated.

5 There is a potential bias created by having the current value of GDP as an explanatory variable when the dependent variable is Currency/GDP, as if there is a random shock in GDP, this will cause a small negative bias to the GDP explanatory variable. The normal way to deal with this is by using an instrumental variable instead of the current value of GDP. As is normal we used a lagged value of GDP itself, but it made virtually no difference.

6 Due to a lack of data availability, we assumed the ratio remained at 2012 levels in 2013.

7 See Changes to National Accounts: Inclusion of Illegal Drugs and Prostitution in the UK National Accounts, 29 May 2014, ONS.

3,989 Reads